CA1180438A - Method and apparatus for lightness imaging - Google Patents

Abstract

ABSTRACT OF THE DISCLOSURE
This invention provides mechanisms and detect the large dynamic range of radiant intensities in the natural environment, that use novel strategies to calculated an approximation of visual properties of objects, and that represent a scene with an image having a specific dynamic range that is optimal for display media such as photography, television and printing.
Photographs and other images are made according to the foregoing mechanisms from lightness fields produced from multiple comparisons between information associated with different segmental areas of an image field. Different comparisons involve different groups of segmental areas, and different groupings have at least one spatial parameter different from other grouping of areas. Comparisons advantageously are made in succession with an ordered sequence of the spatial parameter and employing results of prior comparisons.

Description

BAC!;GROU ~D OF THE IN~.NTION
This invention relates to the art o~ creating and of processing images. It provides a method and apparatus for computing or processing visually 5 perceptible images in terms of a lightness field. A
lightness field is herein defined as the output of a process that uses radiances falling on a light-~etector from an original image to produce a new set of values that correspond to the sensations of lightness produced 10 by the h~man visual system~

Vision science begins with the basic properties of light, for which there are clearly est~blished quantitative concepts tied to physical measures. For example, the radiance from a given regior d~notes 1ux of radiant energy per unit project~d area per unit solid angle. Reflectance of a surface denotes the fraction of incident radiant energy ~of a specified wavelength distributionj reflected by the surface.

Less consensus has developed in characterizing human reaction to light, and it is not surprising that a term such as lightness, which generally refers to a sensation, has at different times been assigned different meanings. Thus, in Webster' s New International Dictionary, Second Edition, lightness was defined as the "state or quality of illumination, or degree of illumination," i.e., a physical measure. In Webster' s l~ew Internatlonal ~ictionary, Third Edition, lightness in addition denotes z sensation, namely, "the 30 at-ribute of object colors by which the o~ject appears to reflect or transmit more or less of the incident light and which varies for surfaee colors from black as a minimu~ to white as a maximum,..".

The later definition recognizes the importance of appearance, i.e. the sensation. Sensation is important hecause the lightness of an object is not necessarily related to the physical quantity of light 5 from the object, either in radiometric or photometric terms. An object will hold its position on a lightne~s scale despite large changes in its intensity~ Much of the difficulty in terminology thus arises because visual sensations characterizing a specific region cannot be 10 directly related to any physical measure of light from that region alonea The term lightness as used in connection with these teachings will have a primary meaning of a visual 15 sensation as produced by biological systems such as human vision~ These lightness sensations are produced by a biological system that takes the radiance at each point in the field of view and yields a lightness ~alue for each point in the field of view, In particular, 20 lightness denotes a visual sensation which ranges from dark to light and which charactexizes image regions in specific conditions. One of ~he more interesting properties of human vision is that the cerebro-re~inal lightness signal processing system is such that the

2~ lightness sensed at any point does not have a simple functional relationship to the radiance a~ that point.
Lightness thus does no~ depend on ~he physical properties of single points or objects in the field;
lightness instead depends on relationships between 30 physical properties of all the points or objects across the field of view. Lightness does not result from point-by-point processing lightness results from processing the entire field.

-3--Lightness can be quantified by employing a technique o visual comparisons. First, one establishes a standard display that include~ reference areas covering the range from minimum to maximum reflectance S in controlled illumination and surround~ One then presents an observer wi~h another area in any viewing condition and asks the observer to select t~e best visual match of that area to a reference area of the standard display. Finally, one takes the reflectance of 10 the chosen reference area and typically applies a monotonic scaling function so that equal increment~ in the resultins lightness num~ers are assigned to equal changes in sensation. Such an approach emphasizes the fact that although lightness is a sensatiQn produced by 15 a human or other biological system, it is a quantifiable entity~ The correspondence in reports from large numbers of observers in numerous experiments of this type shows that these sensations are ~enerated by a repeatable set of physical relationships. Since 20 li~htness depends on the entire image, a physical definition of ligh~ness must incorporate a process which utilizes the entire field of view, The teachin~ herein describes signal 2~ processing systems w~ich generate quanti~ies that correspond to lightness. The quantities, however, are generated by machine signal processing systems rather than biological systems. For c~arity we define a separate term to describe these machine-generated 30 quantities that correspond to li~htness. We have chosen the term "lightness field" as the name of the output of the machine for the selected field of view. This choice emphasizes the fact that a lishtness fleld is derived 31~1 frcm ~ignal processing operations ~ihich involve the field of view, l~is characteristic of lightness field computation distinguishes it from other signal processing strategies that in~olve either single points 5 or local areas of the image~

~ uman vision is remarkable or its ability to generate sensa~ions that correspond to the physical properties of object~ in the field of view regardless of lO ~he radiant intensity and of the wavelength distribution of the light falling on the retina. The wavelength-intensity distribution of the light from an object falling on a ligh~ detector s~ch as a photosensitive element is a function of two independent variables: the 15 illumination at the object and the ability of the object to reflect or transmit light. However, the radiance measurements for any single picture element, i.e. pixel, are not subject to an analysis w~ich identifies the independent contributions of illumination and of object 20 properties.

This invention, on t~e other hand, uses the entire field of view to calculate visual properties of objects substantially independently of the properties of 25 the illuminant. ~sing the en~ire field of view is consiAered essential to a solution of the problem ~hat cannot presently be solved by processing information at individual pixels independently of that at other pixels.

It is difficult for a photograph or like image to accommodate variabilities of lighting conditions, even when care is taken ~o center the limited dynamic ran~e of the image medium on the aynamic ran~e of the --5~
ligh~ being recorded. Consider the light refLectecl from a collection of different colored and textured objects, ranging fro~ the brightest white to the darkes-t ~lack, when special effort is taken to illuminate the 5 collection so that the same intensity of light of the same spectral composition falls on each point in the field of view. ~he dynamic range of the light reflected from thi~ collection of uniformly-illuminated o~jects is signiicantly less than a range of lOO-to-l. The 10 brightest white objects may reflect roughly only 92~ of the ligh~ falling on them, whereas the darkest black velvet objects may reflect roughly at least 3~ of the lignt falling on them. The light reflected rom objects having matte surfaces falls between these extreme values 15 for bright white and for black velvet.

These physical properties of ob~ects limit reflective reproduction media, such as photographie prints and printing, to a like dynamic range, i.eO to a 20 range significantly less than lO0-to-1.

However, the dynamic range of intensities from real life, iOe. from natural images, is far larger than that in this special uniformly-illuminateA experimen~.
25 Natural scenes include sizable variations in the dynamic range of the illumination. First, natural illumination varies both in overall total intensi~y and in local regions because some okjects are shaded by others, Second, the spectral composition of the incident light 30 ma~f Yary dramatically from skylight to sunlight to tungsten light to 1uorescent light. As noted, human vision is remarkable in that it generates image sensations which are nearly indifferent to this extreme var ability of lightin~ cond~ions. These same var;.ations in illumination, however, produce marked and usually detrimental results in conventional image-reprod~cing systems, ~Yhether ~hotographic, televis.ion or 5 printing~

The present invention endeavors to resolve these imaging problems. More particularly, this invention provides mechanisms that detect the large 10 dynamic range of light intensities, that use a novel strategy to calculate approximations of visual properties of the objects in the field of view, an~ that represent the entire image in a limited dynamic range ~hat is optimal for media such as photography, 15 television and printing. A significant feature of the invention accordingly is the calculation of lightness fields that portray large dynamic ranges of the original sce~e in terms of limited dynamic ranges defined by the range of intensities available in various media.

Various photographic defects result from attemptiny to photograph the natural environment "as is". Ordinarily the photographer consciously tries ~o avoid or minimize these defects by the practice of his art. He measures the light coming from the objects in 25 the scene and adjusts the time and the aperture settings so that the exposure will fall on the desired portion of the limited dyn~mic range of the film. ~e artificially illuminates all or part of the scene to compensate for ~on-uniformities in illumination across the scene. He uses color-correcting ilters to ma~ch the spectral proper~ies of the scene to the spectral sensitivity of ~he film. The pho~ographer makes these corrections in 3~

part by estimating the physical properties of the illumination, perhaps with the aid of a light meter. A
television cameraman and his crew follow similar procedures. Further, present-day automatic cameras S determine the lens aperture and the shutter time settinys, but they do not do all that is necessary to correc~ the range of lighting problems found in a natural environment~

The power o the concepts set forth herein can be illustrated by the following practical e~periments d~monstrating advantages realized and realizable in one practice of this invention. The description is of six experiments that emphasize typical common handicaps 15 presently encountered in photographing complex images.
Typical photographic defects result rom the mismatch between the dynamic color range of an original scene and the limited color and inten~ity responses of photographic materials. For the following experiments, 20 a complex original scene is provided in the form of a recorded and displayed television image. This image is in full color and portrays a wide ranse of hues occurriny in varying densities, for example, a woman in a colorful costume against a bright multicolored floral 25 background.

In each experiment a conkrcl image is described which represents the response to each original scene of a conventional pho~osraphic system that does 30 not employ this invention. The first such control image demons~rates the mismatch commonly encountered between the dynamic color range of an original scene and the limited color response _haracteristic o color film.

For example, hiohlights exhibit a degree of levelling and Aesaturation, whereas shadow areas show li~tle evident image detail. In the irst experiment o the inven~ion, the same original scene is sub~ected to 5 ligh~ness analysis by the lightness imaging system defined below and is photographed on a standard photographic medium. Thi 5 first processed image is found to possesx much clearer image detail in shadow ~nd in hi~hlight areas, a better defined range of ~olor 10 values, and improved saturation. To the eye of an observer, the processed ima~e more accurately represents the content of the original scene than does the con~rol image. In producing the processed image in this firs~
experiment, as in the others described below, the only 15 image information available to the lightness imaging system is that which is contained within the original scene itself.

In a second experiment, the same original 20 scene used in the first experim~nt is modified by the superimposition of a ten-to-one illumination gradi n~
from one side of the scene to the other. When this modified scene is photographed, using con~entional techniques to produce a control image, most of the image 25 detail is lost in the darkest portions of the image, or in the brightest portions, and most of the color values are lost. But when this modified scene is analyzed and photographed using the lightness imaging system of this invention, a second processed color print is obtained 30 which is virtually indistinguishable from the first processed i~age described above. The ten-~o-one illumination gradient has disappeared, and the resultant image displays the same saturation, image detail, znd 3~

pleasing dyllamic range as tha~ of the first processed ima~e. ~`urthermore, this second processed image is obtained by the same lightness imaging system operating in the same way and with no ~urther modifications, 5 adjustments, or revised progra~ningO

A third experiment is performed. The original scene is now subjected to different modification representing tungsten illumination of the scene. As a 10 consequence, ~he intensity of the middle-wavelength illuminant i5 only 41% of that of the long-wavelength illuminant, an~ that of the short-wavelength illuminant i S A mere 5~ of that of the long~wavelength illuminant, An ordinary photograph of this modified scene is 15 strongly reddish with few discernible green color values and with practically no visible blue color values, However, when this ~odified scene i~ processed by the lightness imaging system, operating in the ~a~e unmodified way, a third processed color print is 20 obtained which is virtually indistinguishable from the first twoO

Then in a fourth experiment, the original scene is sub~ected to both of the illumination 25 modifications employed in the second and third experiments. Thus, not only are the color values of the entire scene altered by a tungsten illuminant, but the illu~inant varies by a ten-to-one gradient from one side of the original scene to the other. A conventional 30 photograph of this modified scene is strongly reddish with few discernible green color values and practically no blues, and all the image detail appears lost in the darker portion of the illumination gradient. At this ~8-~?4~8 point it should come as no surprise to learn that indeed the fourth processed i~age obtained by the apparatus and met.nod of this lnvention i5 not only essentially free of the imposed modifications, but is substantially 5 i.clentical to the first, second, and third processed images.

~ co~mon problem in photography, differen~
from those already considered, is that of preserving 10 image detail in distinct areas of a scene that has diffe.rent overall levels of ill~ination. Two additional experiments are described with a new original scene that shows a ~ousehold interior in which a person is seated by a window onto a colorful outdoor view.

In a fifth experiment, t~e new original scene is characterized by an eight-fold reduction in the illumination of the view out~ide the window, that is, this modified original scene depi.cts the illumination of an evening. When this evening scene is photographed, ~0 with the same conventional practices previously ~lsed to produce a control image, most of the image detai 1 and color values in the outdoor portion of the scene are lost. But when this modified scene is analyzed and photographed using the lightness imaging sys~em of this 25 invention, a fifth processed color print is obtained in which the scene is accurately represented both inside and outside the window with the same improvements in image quality described for the previous experiments and in which the outdoor view still appears somewhat darker, 30 as is true of the evening setting.

In a sixth experiment, the new original scene is charac~erized by an eight-fold reduction in ~he r~'~3~

--ll--illumination of the indoor scene in front of the window wit~ no reduction in the illumination of the outdoor vie~ behind ~he window. The modified scene now represents a daytime setting with the indoor portion 5 relatively darker than the bright outdoor view, When this daytime setting is photographed, most of the image detail and color values in the indoor porticn are lostO
Bu~t when this modi~ied scene is analyzed and photographed using the lightness imaging system of this 10 invention, a sixth processed color print is obt~ined in which the scene is accurately represented bo~h inside and outside the window with the same improvements in image ~uality described for the previous experiments and in which the indoor scene appears somewhat darker, as is 15 the actual case for a daytime ~e~ting.

Furthermore, the fifth and the sixth processed images are obtained with exactly the same lightness imaging system operating in exactly the same way as for 20 the fir5t ~our processed imagesO

The invention thus advances the art of retinex processing as disclosed in the literature, examples of w~ich are:
U.S. Patent No. 3,553,360 U.S. Patent No. 3,651,252 E.H. Land and J.J. McCann, "Lightness and Retinex Theory", J. O~. Soc., Am., 61, 1-11 (1971~.
EcH. Land, "The Retinex Theory of Colour Vision", Proc. oyal Inst~ of Gr. Brit., 47 (1974).
J.J. McCann, S.PO ~cKee and T.H. Taylor, "Quantitative Studi~s in Retinex Theory", Visl~n Research, 16, 445-458 (1976).

Other publications in the imaging art are the article by T.G. Stockham, Jr., "Image Processing in the Context of a Visual Model", Proceedings o the IFEEI
Vol. 60, No. 7, July 1972, pages 828 through 842; the 5 article by David Marr, "The Computation of Lightness by the Primate Retina", Vislon _esearch, Vol. 14, pages 1377 through 1388; and the article b~ Oliver D.
Faugeras, "Digital Color Image Processing Within the ~ramework of a Human Visual Model", IEEE Transactions on 10 ~coustics, Speech, and Signal Processin~, Vol. ASSP27, No. 4, August 1979, pages 380-393. This invention employs techniques which differ significantly from the image processing which these articles discuss.

Objects of this invention, and advantages which it brings to the art of imaging, include attaining lightness imaging with fewer signal processing steps or computations in considerably less time than previously available.

A further object is to provide a method and apparatus for lightness imaging applicable on a practical basis to numerous image processing and numerous image creating instances.

Another object of the invention is to provide a method and apparatus for providing an image, termed a lightness image, which represents a scene in a limited dynamic range that is optimal or display media such as photography, television and printing.

It is also an object to provide image processing that uses informati3n acquired at one segmen.al a-ea of an image in evaluating information acquired at other segmental areas in a learning-like manner that attains a desired lightness field in relatively small time and with relatively few proce~sing steps.

It is also an object to provide a method and apparatus of the above character suited for commercial application.

Other objects of the invention will in part be obvious and will in part appear hereinafter.

SUM~RY ~F THE INVENTION
The practice of the invention enables one to 15 produce images in a way that is analogous to human vision because it represents in a limited dynamic range the much larger dynamic range of radiances found in the natural environment. Media using lightness fields produced in this way can have a far greater vi~ual 20 fidelity than previously available on a repea~able basis. Furt~er, the media images can be essentially free of defec~s such as illumination artifacts, color imbalance, and other spectral mismatches.

In accordance with the invention an image is produced from multiple comparisons between radiance information at different locations of an image ield.
The different comparisons are made ~etween different groupings of location~, and at least some groupings 30 involve loca~ions characteri zed by a spatial parameter different from that of other groupings~

3~

In further accordance with the invention, information is provided, for example by viewing a scene with an optlcal detector array, identi~ying the radiances associated with arrayed sections of an imaye 5 field. Mul~iple measures are made of transitions in the radiance information between each segmental area of the image fie]d and other such areas of the field. The several measures involve groupings of each area with othe~ areas in a way such that the areas of different 10 ~roupings differ in at least one spatial parameter. The different grouped areas cover different sectors of ~he viewing field and consequently can be areas separated by different distances, areas separated along different directions, and areas of different sizes. These 15 measures - produced in response to radiance information and from such differently grouped segmental areas of the viewing field -- are combined to provide the desired lightness information for the entire image field.

The measures of iransi~ion in radiance information preferably are determined for different groupings of segmental ~reas according to a sequential ordering which proceeds, for example, from groupings of maximu~ spatial parameter to groupings of progressively 25 smaller spatial parameter. This sequential ordering has been found to diminish the level of unwanted artifacts in the processed image.

A eature of the invention ix that it produces 30 an image from multiple comparisons between radiance information a~ different locations in an image field by proceeding on a field-by~field basis~ Each iteration of the process, in ~he illustrated embodiments, maXes new co~par,sons for essentially all locatlons in the entire field~ This is in contrast to a prior prac~ice in which each basic operation provides a new comparison for only one location.

Another feature of the invention is that it produces an image from multiple compari sons between radiance information a~. differen~ locations in the image ield by also using a combined reset-pooling technique.
The technique enables image information accumulated for 10 each location to be used in evaluating image information for other interacti ng loc~tions.

It is also a feature of the invention that ~he multiple image producing radiance comparisons combine 15 local image information with information from more distant parts of the imageO This local global computation attains a lightness field of desir~d quality with rela~ively few computations and in relatively shork time.

Ye~ another fea~ure of the invention is tha~
i~ combines the foregoing field-by-ield computations, information accumulation by reset-pooling techniques, and local-global calculations to accomplish in real time all caleulations for imaging complex natural scenes.

In the embodiments described below, the ~roupings of se~mental areas are pairings, i.e. each involves two segmental areas or picture element~.
Further, each measure is made wi~h two pixels, i.e.
picture elements, that are identical in size and in 30 shape. The tw~ pixels in an illustrated pairin~, however, diEfer in a spatial parameter from other paired pixels.
In one instance, the pixels of one pair are of diEferent size from the pixels of another pair. In ano-ther instance, one pclir involves pixels separated by a distance d:ifEerent from the separa-tion between other paired pixels, and/or one pair involves pixels separated along a direction diEEerent from the separa-tion direc-tion of other paired pixels.
Also in the embodiments below, each measure of the rad-iance information at the paired pixels is determined with a tran-sition measure, e.g. with the ratio of the radiance informationassociated with two paired pixels~ This ratio is multiplied by a dimensionless value previously assigned to or determined for one pixel of the pair, i.e. for the divisor pixel of the ratio.
The resultant product is reset with reference to a selected limit.
The reset ratio product can serve as the desired comparison mea-sure for a given pairing of pixels.
However, in a preferred practice of the invention, the reset ratio product is pooled by combining it with a dimensionless lightness-related value previously assigned to or determined for the other pixel of the pair. The resultant from this operation is the desired measure for that pair of pixels. This further step is desirable because it increases the rate at which radiance information is accumulated from different pixels. It hence decrea-ses the number of computational steps needed to attain a lightness image.
More particularly, each radiance-transition measure for '~' U~3~
a pix~l contains information regarding the lightness (3enerating proper-ty to be provicled a-t -that location in -the resultant imaye.
An objective for lightness imaging in accordance wi-th the inven-tion is to compare -t,he radiance informat:ion ~or each plxel with that: of subs-tantially all others and there~by determine the ligh-t-ness proper-ty which each pixe] has in the complete image field.
The ou-tput of -this process for all locations is the lightness field. The repetitive replacement of the reset ratio produc-t information for one pixel in each pair accumulates this informa-tion relatively slowly. The rate of information accumulationis increased significantly by combining the reset ratio produc-t from one iteration with the measure previously determined fox or assigned to one pixel in each pair, and using the combined measure for -the next iteration, i.e. in the next pairing ot tha-t pixel.
To this end, a preferred embodimen-t of the invention provides that each reset ratio product measure of a transition in radiance informa-tion be combined according to an arithmetlc averaging function with the prior measure assigned to -the other pixel of that pair. This pooling of information provides a geo-metric increase in comparisons of the radiance information of different pixels from each measuring iteration -to the next meas-uring iteration. Consequently, it markedly decreases the number of iterations required to produce an image, and thereby con-tri-butes to a new fast rate for image produc-tion. The combination of this pooling of reset ratio product measures with a selec-ted , . . .

~3q ~3~3 sequ(~ntial ordering of pixel pairings Eor effecting the measure-ment; yields compound advalltages in accumulclting inEormation for the creation of lightness images.
In a preferred prac-tice of -the pooling embodiment oE
this invention, the imaging process accumulates information in a learning-like manner. The information at each pixel, at the end of any iteration, is the combined measure of all the reset ratio products that have so far reported to that pixel. Thus the number of pixels affecting the reset ratio product can be equal to two raised to the power of the number of iterations.
If there are eighteen iterations, then the number of interactions is equal to two to the eighteenth power. By way of contrast, a prior process brings to each pixel a reset ratio product carry-ing the information from a number oE pixels equal to only the number of iterations.
A mode of operation intermediate to the iterative repla-cing of reset ratio product measures and the pooling of such measures, is the averaging of reset ratio product measures from different sets of pairings. Another operating mode involves deter-mining reset ratio product measures for a set of pixel pairings,and using the resultant measure in determining further measures with a successive set of pixel pairings with a different magnifi-cation of the image field.
These and other features of the invention are described below with reference to different image-producing embodiments.
One embodiment involves pairs of pixels of identical size and `q~? ~
-18a-configuration. It employs a sequence of pixel pairings ordered both with successively smaller pixel spacings and with differen-t directions of pixel separation. The processing of ~ ~¢~ 3-~

--lq--radlance information in these embodiments preferably proceeds on a field-by~field basis. That is, the iterations for any one pixel or other segrnental area of the image field occur essentially in step with 5 iterations for other pixels o the fiel~. The steps wlthin each iteration CAn occur either time seq~entially ~r in parallel for the different pixels. Radiance information is compared in the field-by-field basis by shifting a pattern of radiance information and comparing 10 it to the unshifted pattern. The shifting is illustrated both on a time basis, e.g. by use of del~y lines, and on a spatial basis, e.g. by the use of known scroll techniquesO

Other embodiments involve pairs of pixels that represent areas of the image field different in size from those which other pairings represent. The measurements of radiance transition in either instance can proceed in different sequences to attain replaced 20 reset ratio produc~ measures, to pool the measures from one iteration or set of iterations to the next, or to average the measures from independent sets of iterations.

These and other image-producing features of the invention may have wide application. ~pecific instances include photographic processing, e.g. in a camera as well as in print making, and television signal processin~, e.g. in broadcast equipment as well as in a 30 receiver. Other instances include devices, for example laser scanning devices, used in graphic arts to scan documents and generate corresponding printing plates.
In these applications, as the above-described experiments illustrate, the invention enhances the overall light-ness quality o~ the displayed image. In addition, it corrects for a varie-ty of visually detrimental fac-tors including illumina-tion deficiencies and materia] and equipment limitations in repro-ducing wide bandwidth and in-tensity ranges.
Fur-ther applications of the invention can produce displays in instances where a scene is examined non-optically, and even where there is no original scene. Examples of the former include the creation of an image display of an object examined with sonar techniques, with infrared techniques, or with radar techniques.
Instances of the latter app]ication include displays produced with computerized axial tomography (CAT) scanners and with other nuclear medicine instruments. All these applications of light-ness imaging in accordance with the invention can produce images that are consistently perceived as more satisfactory than those heretofore available. Those skilled in the art will realize from these teachings that the practice of the invention can also pro-duce images with all amnner of specially contrived lightness effects.
The invention accordingly comprises the several steps and the relation oE one or more of such steps with respect to each of the others,and the apparatus embodying features of con-struction, combinations of elements, and arrangemen-ts of parts adapted to effect such steps, all as explained in the following detailed disclosure, and the scope of the invention is indicated in the claims.
Some, but not all, of the various aspects of the invention are summarized in the following statements of invention:

~"i i ..,-, ..

-20a-According to one aspect the invention provides apparatus for producing an image of a subject which comprises A. means for detec-ting radiance ratios between di:~ferent areas o:E sa:id subjec-t and producing a first lightness-dete:rmining quantity in response to each such ratio, B. means for eEfecting said ratio detection for each area of said subject a multiple numbex oE times with other areas of said subject which are at different locations on said subject relative -to that area, C. means for combining each first lightness-determining quantity with a second lightness-determining quantity associated with one area in that ratio andreplacing the second lightness-determining quantity associated with another area in that ra-tio in response thereto, and D. means fox producing an image of the subject in which the lightness of the respective image areas is determined by the last replacement values of said second lightness-determining quantities.
According to another aspect the invention provides image processing apparatus for determining information corresponding to image lightness in response to radiance-identiEying informa-tion for a selected image field, said apparatus having the improve-ment comprising A. means for representing the radiance-identifying information for each of selected segmental areas of the viewing field, B. means for determining a selected comparison measure between said identifying information for each segmental area and said information for another segmental area, and for determining therefrom and from a previously-determined lightness-identifying quantity for each latter segmental area a newly-determined light--,j i ~

3~3 -20b-ness-identifying quantity for each former segmental area, C. means for effecting a selected multiple of said determinations sequen-tial:ly and between segmental areas that correspond to differen-tly-spaced locations in said field of view, and D. means for producing said lightness-information for said image :Eield in response -to said multiple determinations.
According to a further aspect the invention provides in lightness-imaging apparatus having (i) means for providing infor-mation identifying optical radiance associated with each arrayed section of a selected image field, (ii) means for selectively pairing segmental areas of said image field a selected number of times, each said pairing being of segmental areas of identical configuration and size, (iii) means for providing, for each pairing of segmental areas, at least one comparative measure of said rad-iance information at the paired areas, and (iv) means for reset-ting each said measure with reference to a selected limit condi-tion, the improvement comprising means for determining image lightness for each arrayed section of the image field in response to a plurality of said reset measures, at least some of which are provided for pairings which differ from one another in at least one spatial parameterO
According to yet another aspect the invention provides lightness-imaging apparatus having means for providing information identifying op-tical radiance associated with arrayed sections of a selected image field, said apparatus further comprising A. means for pairing identically configured and sized segmental areas of said viewing field di.fferently a number of times and , . ..

3~3 -20c-for providing a multiple of sets of said different pairings, each said set involving areas of a size diferent from other sets, B. m~ans for providing, for each palr.ing of segmental areas, a comparative measure of said radiance informati.on at the paired areas, C. means for rese-tting each said measure with reference to a selec-ted limit, and D. means for determining image llghtness for each arrayed section of the image field in response to a plurality of said reset measures.
According to still another aspect the inven-tion provides a method for producing an image of a subject comprising the steps of A. detecting radiance ratios between different areas of said subject and producing a first lightness-determining quantity in response to each such ratio, B. effecting said ratio detection for each area of said subjec-t a multiple number of times with other areas of said subject which are at different locations on said subject relative to that area, C. combining each first light-ness-determining quantity with a second lightness-determining quantity associated with one area in that ratio and replacing the second lightness-determining quantity associated with another area in that ratio in response thereto, and D. producing an image of the subject in which the lightness of the respective image areas is determined by the last replacement values of said second lightness-determining quantities.
According to another aspect the invention provides a lightness-imaging method in which information is provided identi-fying optical radiance associated with arrayed sections of a selec-.

~20d-ted image field, said method further comprising the steps of A. selectively grouping segmen-tal areas of said imageEield a selec-ted number of times, di:Eferen-t ones of at least some o.E said group-ings involving areas having at least one spati.al parame-ter diE-ferent from other groupings o:E areas, ~. providing, :Eo.r each group-ing of segmental areas, at least one measure of visually signi:Ei-cant -transition in said radiance information between areas of that grouping, said measures being with reference to a selected lightness condition, and C. determining i.mage lightness for each arrayed section of the image field in response to a plurality of said measures, at least some of which are provided for group-ings which differ from one another in a-t least one spatial para-meter selected from the parameters of distance, direction and size.
According to still another aspect the invention provides image processing apparatus comprising A. means for receiving in-formation responsive to the radiance values defining an image field, and B. means for deriving from said information a lightness field containi.ng final. lightness values for predetermined segmen-tal areas of said image field, said final lightness value derivingmeans establishing initial lightness values for all areas of said image field and sequentially performing a selected number of pro-cess steps for said image field, in each step of which process selected areas of said image field are selectively paired with different areas of said image field and in successive steps of which p-ocess such pairings of areas differ from other pairings -20e-in at least one spatial parameter accordiny to a predetermined sequence~ and in each of which steps such paired areas are compared to establish a new lightness value for each said selec-ted area as a :Eunction of the ra-tio of its radiance value -to that of the diEferent a:rea with whictl it is paired and as a functlon of light-ness values established for such paired areas in a preceding pro-cess step~ and wherein said final lightness value for each said segmenta:L area comprises an effective comparison of information responsive to its radiance value to information responsive -to the radiance value from substantially all other areas of said image field without a direct comparison to each of said other segmental areas.

f" i, '438 BRIEF DESCRIPTION O~ DRAWINGS
__ For a fuller understanding of the nature and objects of the invention, reference should be made to the following detailed description and the accompanying 5 drawings, in ~lich:

FIG~RES lA, lB, lC and lD are diagramma~ic illus~rations of an information pooling feature o the invention;
FIGURE 2 is a vector pattern partially illustrating a progression of spatial parameters fcr pairing pixels in accordance with one practice of the invention' FIGURE 3 is a ~lock schematic represen~ation of an image-producing system embodying features of the invention;

FIGURE 4 is a flow chart illustrating a 20 practice of the invention with the system of FIGURE 3;

FIGURE 5 shows five look up table functions for use in the system of FIGURE 3;

FIGURE5 6A, 6~, 6C and 6D are schematic drawings of one image processor of the system of FIGURE
3 illustrating successive operating stages' FIGURE5 7A, 7B, 7C and 7D are diagrammatic 30 representations of a portion of an image processor of FIG~RE 3 and illustrate one implementation of wraparound insulation in accordance with the invention, 3~

FIG~RE 8 shows two look up table functions for the practice of wraparound insulation as illustrated in ~IGURES 7A~D;

FIGI~RE 9 is a block schematic diagram o~
another embodiment of an image processor in accordance ~ith the invention' FIG~RE lO is a diagrammatic representation of lO a mapping o information in a memory for practice of features of the invention, FIGURE ll is a block schematic representation of another image-producing system embodying features of 15 the invention, and FIGURE 12 is a diagramma~ic representation of a self-developing camera in accordance with the invention.

. _ _ _ _ The processing of ~his invention determines a measure of all transitions in the radiance associated with each o different groups of areas or pixels in an image field. One practice involves groupings of two 25 pixels, i,e. pairs of pixels, and determines the desired measure as the product of ~i) the ratio of the radiances at the paired pixels and (ii) an existing measure initially assigned to or previously determined for the pi~el associated with the denominator of the 30 ratio~ Numbers proportional to logarithms of the relevant measures are typically employed, so that an adder can perform the computations: the 105 of the ratio is the sum of the log of one radiance plus the negative log of the other radiance, and the log of the produc-t is the sum of the log of the exis-ting measure ancl the log oE -the ratio.
The resultant ratio-product, also termed an in-termediate product, is referenced to a selected ligh-tness condition. A pre-ferred practiceis to reset each resultant which exceeds the value equivalent to maximum image lightness to that value. This rese-t operation establishes those specific locations or pixels in the scene which are lightness reference points for subsequent opera-tions. The above-noted U.S. Patents Nos. 3,553,360 and 3,651,252 and the above-noted article of McCann, McKee and Taylor discuss prior practices regarding the foregoing ratio, product and reset operations.
The reset intermediate product determined with each such measuring iteration can be used as the existing measure for the next iteration. However, the invention attains a decrease in the numher of iterations required to produce a high quality image by averaging the reset product with the existing measure at the pixel which is associated with the numerator of the ratio for that pair. The resultant averaged product is the transition de-sired measure for that pixel, as determined with that pair of pixels at this iterative stage in the sequence. This averaged product also replaces the prior measure for that pixel and is used in the next measuring iteration. This averaging opera-tion increases significantly the information which the resultan-t aver-aged product contains, because in each iteration information from nearly every pixel contributes to two resultant products.

The objective for lightness imaging in this way is to determine a comparison measure Eor each pixel relative to every other pixel, i.e. -to compare the racliance oE each pixel with that of every other pixel by subjecting selected xadiance ratios to the foregoing product, reset and averaging opera-tions. In prin-ciple, there are many practices of these comparison-measuring operations that ca~l yield -the desired lightness information for creating a high quality image. ~owever, in addition to using an averaged product rather than simply the rese-t product as just discussed, the invention minimizes computa-tion by using multiple groupings of pixels, each with a spatial parameter different from other groupings or at least other sets of groupings. One specific example is to use pairs of pixels, each of which differs from other pairs in the spatial separation between the paired pixels, and/or in the direction of the spacing between paired pixels.
Another specific example is to compare with one pairing pixels different in size from those compared in other pairings.
Another feature of this invention is the realization that an image can be broken down into multiple groupings of pixels in a manner such that both local and global interac-tions, e.g.
comparative measures with proximate locations and with remote locations, can be calculated at an extremely rapid rate. Dif-ferent segmental areas are compared so that long-distance spatial interactions can be computed in much less time than with proces-ses involving only contiguous pixels. By transforming the image into a representation that contains comparatively few segmental ?,', 3~

areas, locations which are separatec1 by long distances in the original represen-tation of the image are now compara-tively close.
This strategy loses resolution, or in-troc1uces errors, due to the coarseness of -the analysis. Never-theless, this ability to perEorm long clis-tance calcula-tions in few iterations yields enormous reduc-tions in the number of iterations and correspondingly in the pro-cessing time. Further, the problems regarding resolu-tion and characteristic errors are overcome by a judicious choice of segmen-tal areas in subsequent stages of the calculation. A corollary feature is the provision for a suhdivision of the image into var-ious segmental areas for separate calculations o-f various compon-ents of a lightness field, followed by a recombina-tion of the processed inEormation into a lightness field that is influenced by each characteristic set of segmental areas.
Still another iteration-saving feature is to order the comparison-measuring iterations according to the magnitude of the pixel spacings or other spatial parameter. These features, together with others described hereinafter, are remarkably effec-tive in attaining a close approximation to the foregoing objec-tive on a real-time basis.
FIGURE 1 illustrates the effectiveness of the foregoing features in inter-comparing various elemen-ts of an image field with a limited number of operating iterations. FIGURE 1 is arran-ged in four columns, i.e. FIGURE lA, lB, lC and lD, and in three rows. Each column from left to right represents a successive comparison-measuring iteration. The -top row shows the 3~

same sixteen arrayed sections A, B, C...o and X of an image field for each of the four iterations. The sixteen sections of the field, or sixteen picture elements, i.e. pixels, are for clarity shown well 5 spaced apart, but in actuality adjacen~ ones are often contiguous or nearly contiguous. The middle row of FIGURE l shows ~hose pixels of the top row which, at the end of each iteration, have contributed radiance information to a measure made with the pixel (X)~ The lO lower row shows a vector representation of the magnitude and direction of the spacing between paired pixels.

The first illustrated iteration, FIGURE lA, shows the aforesaid measure oS radiance transition 15 between each pixel and the pixel two locations to the left of it. Directing attention to pixel (X), the radiance there is compared with the radiance of pixel (I). The specific computation forms the ratio of the radiance at pixel (X) to that at pixel (I~ and 20 multiplies the ratio by the existing measure at pixel (I) to form an interme~iate product. The interme?liate product is reset and averaged with the existing measure at pixel (X). This computation yields the desired averaged product at pixel (X). It is a function of the 25 radiances at pixels (X) and (I) and of the previously existing products at these two pixels, '~e second row in FIGURE lA accordingly shows these two pixels. The third row designates with a single horizontal vector the direction from which pixel (X) received radiance 30 information. Simultaneous with these measuring steps, every other pixel (A) through tO) in the upper column of FIGURE lA receives radiance information from the pixel two locations t~ the leLt of it, neglecting for the 3~

moment pixels in the left half of the image field for t~iere are no pixels two locations to the left of them.

~'IGURE l~ illustrates that the nex~ iteration S measures a radiance ~ransition ~etween each pixel and the pixel two locations above it. The radiance at pixel (X) is accordingly compared with the radiance at pixel ~C), and the ratio is multiplied ~y the existing product determined for pixel (C~ in the first iteration. After lO resetting and averaging with the ex~sting product just determine~ for pixel (X), the result is a new averaged product for pixel ~X).

This averaged product is a function of both 15 the radiances at pixels (X) and (C), and the existing products, i.e~ averaged products, determined for each of these locations in the first iteration. Hence the new avera~ed produ~t assigned to pixel (X), after the second iteration, is a ~unction of the radiance information at 20 pixels (X), (I), (C) and (A). The middle row in FIGUR~
lB depicts this multiple feeding or contribution of radiance informa~ion to pixel (X). The third row o FIGUR~ lB represents the second iteration pairing as a vector which starts from the terminus of the first 25 iteration vector. This cumula~iYe vector representation reflects the progressive accumulation of information at pixel (X), that is, the measure at pixel tx) has a history from the prior iterations. Again, information or every other pixel in the illustrated sixteen-pixel 30 array is processed during the second iteration in the same marner as just described for pixel (X).

The third iteratior., shown in FIGUP~E lC, pairs each pixel ~ith the pixel one location to the right. It 3~3 hence relates pixels (X) and (K). The existing measure at pixel (~) is an averaged product with a history of two iterations and hence is a function of -the radiances at pixels (J), (B) and (D).
~ccordingly, the new averaged product at pixel (X) is a Eunction oE the radiances at eight pixe]s/ as tile second column in FIGURE
lC depicts. The vector represen-tation of the pixel jump for this third iteration appears in -the lower row of FIGURE lC.
The illustrated fourth iteration pairs each pixel with -the one located one uni-t below it, FIGURE lD. The information for pixel (X) is hence compared with that at pixel (N). The aver-aged product for each of these locations has a di-fferent eight-pixel history, as FIGURE lC shows in the middle row for pixel (X), developed over the three prior iterations. Consequently, the resultant new averaged product at pixel (X) is a function o-E the initial radiances at each of -the other fifteen pixels, as the simplified FIGURE lD depicts in the second row. (The sim-plification in the second row of FIGURE lD is that only some prior interactions are shown.) The pairing vector between pixel (N) and pixel (X) Is added to the prior vectors to yield the vector representation which appears in the lower row of FIGURE lD. Note that successive vectors extend at right angles to one another.
Note also -that the successive vectors, Erom iteration one to four, are either of the same or lesser size.
In this manner, four relatively simple iterations compare the radiance at each pixel with that of fifteen others. Expressed mathematically, the process is such that after (N) steps the rad-iance at each pixel is compared with those at (2N - 1) other pixels.

-2~-FIGURE 2 shows twe' ve steps of an actual eighteen step vector pattern with which the invention has been successfully practiced with an image field having a (512) x ~Sl~) array of pixels~ For clarity of 5 illustration, FIGURE 2 omits t~e two initiaï largest steps and the final four smallest steps, The illustrated pattern involves an ordered progression of i~erations commencing with two successive s~eps of ( 56) pixels each (not shown in FIGURE 2), followed by the two 10 steps shown of (128) pixels each, and proceeding as shown with two steps of sixty-four pixels each, two of thirty-two pixels each, two of sixteen pixels each, two of eight pixels each, and two of four pixels each.
These are followed by four steps not shown: two of two 15 pixels each and two of one pixel each. The spacings between the pixels of successive pairings thus progressively decrease in length. The directions of the spacings also progressively change, e.g. successive directions in the illustrated pattern (including the six 20 steps not shown in FIGURE 2) are perpendicular to one another in clockwise order. The eighteen comparison-measuring iterations with this ordered succession of pairing steps yield an averaged product for each pixel, aside from those which are to be compared with locations 25 beyond a boundary of the image field, which is the resu~t of comparing the radiance there with those at over a quarter-million other pixels. The final averaged products can be used to produce a photographic image having the significantly improved lightness qualities 30 described in the foregoing examples, The image processing described with reference to FIGURES 1 and 2 incorporates several features. ~ne ,~.s .
. . , ,;~ . ."~

V9~3~

is that it uses field-by-field computations so that each itera-tion ca:lcula-tes a new rneasure for each pixel. The example in FIGURE 1 explicitly describes the process -that calculates the measure for pixel (X). That process, however, is a one-pixel part of multi-pixel Eield-by-field computations that yield, a-t each iteration, a new measure for each pixel in the field. Fur-ther, each iteration after the first one brings every pixel -the informa-tion already accumulated by another pixel, thus producing a geo-metric growth of interactions.
The process next uses a reset-pooling technique that calculates, for each pixel, a combined measure that is an optimal lightness field value for that pixel. This resultant can be at-tained by averaging each new reset ratio product value with the previously assigned or determined value. Each ratio product value is reset before it is averaged with the previously determined value for that pixel; the reset mechanism removes from the combined measure those ratio product values that are known to report light-ness field values higher than the particular imaging medium can display. The presence of this severely non-linear resetting oper-ation distinguishes this image processing from others that simply compare radiances to average values computed over portions of the image or over the entire image.
In addition to this reset-pooling technique, which accum-ulates information in a learning-li~e manner, the processing in-corporates the noted local-global calculations that reduce the total number of calculations to gain the same information, Local-global comp~tatlon~ are preferred to attain a satisfactory lie3htness field calculation whlch does not ignore any poItion of the image field~ The global or long-di~tance 5 interactions provide each portion of the image with the correct relationship to distant parts o-E the image.
~ocal interact:ions are important as well, because they provide high resolution information which reliably relates nearby points to each other. The composite lO technique just described processes the entire image using ~ield-by-field computations to sample information for long distance interactions. It then processes the entire image again using slightly shorter distance interactions, and combines the results with the 15 rese~-pooling technique. The process continues in this manner until it examines the image ~ith single pixel resolution.

Each processing iteration illustrated in 20 FIGURE 1 preferably involves at least four steps, i.e, ratio, product, reset and average, with a field of radiance information like the field associated with the

(4) x (4) array of pixels in FIGURE l. ~ach item of information in the field stems from the radiance at a 25 particular location, i.e. pixel, in the two-dimensional image field and hence can be identified by labeling with the coordinates of that pixel.

The fir~t step of each iteration is to pair 30 the pixels of the image field and to compute for each pair the ratio of the radiance values at the paired pixels. The ratio is a measure which compares the radiance at one pixel, termed a source pixel, te the ~v~

~ 32-radiance at the paired pixel, termed a comparison pixel~
~hen the radi~nce information is loyarithmic, the log o~
this ratio can be calculated as the arithmetic difference between the logs of the paired radiance

5 values. 'Fhus where r(o,o) represents the ra~iance information at the origin pixel a~d r~x,y) represents the radiance value at the comparison pixel, the ratio operation which the first step performs can be represented by the alsebraic expr~ssion:

r(x,y) log - = log r(x,y) ~- log r(o,o) (1) r(o,o) The second step in the processing iteration is to multiply this ratio by a product previously assigned 5 to or determined for the origin pixel of each pair, ~o deter~ine an intermediate product for the comparison pixel. The multipllcation can be performed by the addi~ion of logarithms of the numbers. ~ence, the algebraic expression for the log of the intermediate 20 product at the comparison pixel is log ip(x,y) = log op(o,o) + log r(x,y) - log r(o,o) (2) where:
5 loy ip(x,y) is the log of the intermediate prcduct at the comparison pixel of the pair; and log op(o,o) is the log of the old product previously existing, i.e. old product, at the origin pixel.

The system is initialized for the first iteration by assigning an old produc-t to each pixel. The inikializing product preferably corresponds to an extreme optical condition such as total black-ness or full whiteness. The illustrated embodiment initializes with the latter, which corresponds -to an initializ:ing value oE
unity.
In each processing iteration the third step ta~es inter-mediate products which are greater than unity and resets them to unity. A star (*), designates a reset intermediate product in the fol]owing equations (3) and (~) The fourth processing step combines reset intermediate products to form, for each comparison pixel, a new averaged pro-duct to be used in the next iteration~ As described with refer-ence to FIGURE 1, it has been found preferable to produce the new averaged product as the geometric mean of the existing or old averaged product at the comparison pixel in each pairing and the new value of the reset intermediate product that pixel formed as set forth in equation (2). Where the term "log op(x,y)" is the logarithm of the old averaged product at the comparison pixel, -the new product (np) at that location is thus calculated as log np(x,y) = [log op(x~y) + log ip*(x,y)]/2 ~3) and is defined as ~ ,, ." ".,, 3~3 log np(x,y)=

1/2{[1Oy op(x,y)~lo~ op(o,o)-~lcg r(x,y) lcg r(0,0)]*} (4) This log of the new averaged prod~ct at the comparison loeation (x,y) is used as the ]og of the old product term for that location in the next iteration~

The invention can be practiced without the ourth prccessing step. That practice uses the reset interme~iate pr~duct as the comparison measure 10 assigned to the comparison pixel for the next i~eration.
The preerred practice, however, includes the fcurth step, which improves both the ef iciency of the process and the quality of the results.

15 An Image Producing System FIGURE 3 shows a full-color image-producing system which implements the foregoing lightness-imaging techniques. The system has an input stage 12 that develops information identifying the optical radiance of 20 a field of view to be displayed~ The illustrated input stage is a camera, e,g. photographic or television, and has an optical detector 14 that receives the light energy from a lield of view or other original image 15 via a lens system la. The detector 14 is typically a 25 multi-element array of photosensitive elements, each of which produces an electrical signal in response to the light energy incident on it. The detector response preferably is proportional to the logarithm of the light energy to facilitate subsequent signal processing.
30 Examples o such a detector are an array of charge .~, ~ ..

coupled devices, i.e. a CCD array, or a charge induction (CI~) array.

The illpUt stage 12 ~hus applies electrical signals proportional to the logarithm of radiance to 5 each of three identical image processors 20, 22 and 24, one for each of the red, blue and greell color bands as conven~ional in full color electronic image processing.
Each processor 20, 22 and 24 processes the radiance-identifying signals in a single color band to develop 10 signals identifying the image lightness property for that color band at each point in the imag~ field, e.g.
in the field of view of the lens system 18.

One processor ~0, shown in further detail, has 15 a flrst refresh memory channel 26 which has a refresh memory 28, a scroll device 30, and a look up table 32.
The refresh memory can be a random access memory to store the field of image information from the input stage. The illustrated input stage detector has a (512) x (512) CCD array and the refresh memory has capaci~y to store the radiance inormation for each detector element as a by~e of eight bits. In a typical representation of signals, the maximum possible response is assigned to the level 255, and 0.01% response is assigned to the 25 level 0, with logarithmic signal increments evenly assigne~ to the levels between. The scroll device 30 can displace the field of image information from the refresh memory independently along both the (X) and the (Y) directions by a number of specified coordinates, It 30 typically employs a shift register and a memory to recall a sequence of scroll positions. The look up table 32 employs a random access memory that is 3~
addressed by each information byte being processed to provide a real -time txansformation. The table 32 provl.des a nega-tive trans-formation function. The term "negat~ve" denotes an arithme-tic function having a slope of (-1). The eE:Eect of processing a field of image information wi-th such a pol.arlty inverting :Eunc-tion is to convert between a positive :image and a negative image.
THE FIGURE 3 system has a second refresh channel 34 with a refresh memory 36, a scroll device 38, and a look up table 40.
A third refresh channel 42 has a refresh memory 44 and a look up table 46, but requires no scroll device. An adder 48 is connec-ted with the look up tables 32, 40, 46 from each of the three channels 26, 34 and 42. It sums any active inputs and applies the resul-tant to an adder output line 50. A feedback connect~on 52 applies the output from -the adder, by way of a look up table 54, selectively to the input oi the second channel memory 36 and third channel memory 44.
The system output stages are illustrated as a nonlinear color masking stage 58, an exposure control stage 60, and a display stage 62. A program control unit 56, typically including a pro-grammable processor and connected with each processor 20, 22 and24 and with each stage 53, 60 and 62 as, illustrated, controls the system operati.on. The color masking stage 58 provides a color masking operation which accentuates the color of each area of the image field and compensates for color desaturation both in the input stage detector 14 and in the display stage 62. By way of example, a general purpose of col.or masking in photography is to correct for differences between ideal ~g dyes and dyes real-;zable in ~ctual photographic systems.
The literature regarding such color masking includes:
Clulow, F. W., Color Its Principles And Applications, publ.ished by Morgan ~ Morgan in New York, 1972, pages 5 157-159 and 172-179; and Hunt, R.W.G., The Reproduction Of Color, published by Wiley in London, 1967, pages 233-263 and 3~3-416. The color mask stage 58 can also correct for the limited color response and limited color transmission cap~bilities in the various stages and 10 elem~nts of the system. The color mask stage 5~ thus typically provides color correction, color enhancement and color masking to optimize the output signals from the image processors 20, 22 and 24 for subsequent color display. It can also provide an antilog conversion 15 function unless t~at function has been provided in adder 48.

The exposure control stage 60 transforms the processed and color masked lightness-identifying signals 2~ into the format which matches the display stage 62. In a television-imaging system, the display element 62 typically includes a cathode ray tube television display such as a video monitor, whereas in a photographic camera system this element typically includes a 25 lig~t-emitting diode (LED) array arranged to expose photographic film.

The foregoing elements of the ~IGURE 3 system can be conventional devices known to those skilled in 30 the artt including the arts of computer graphics, electronic image processors, and image computers. By way of ex~mple, the International Imaging Systems division of Stanford Technology Corporation markets an 3~

image computer which employs elements sui-table for -the image-producing system of FIGURE 3.
~y~em _pera ion __ FIGURE 4 is a flow chart of the Eoregoing four step ite:r-ation applied to ti~e F~IG[lRE 3 processor 20. This processor is typical of the other processors 22 and 24, or the three image processors can be identical in construction and in operation, and operate independently of one another except as the program control unit 56 imposes simultaneous or other -time-coordi,nated operation. The three refresh memories 28l 36 and 44 of the pro-cessor 20 are assigned different rows of the flow chart, and suc-cessive operations are shown at different positions along the chart starting with the initial conditions at -the left and progres-sing t~, the right. The processor is ini.tialized by storing ori-ginal image radiance information received from the input stage 12 in the first channel refresh memory 28. The initial contents of the second channel memory 36 are not significant. The third channel memor~ 4/~ is initialized w;.th reset ratio produc-t values corresponding to 100~ reflectance, as discussed above.
The flow chart of FIGURE 4 is described further together with FIGURES 6A, 6B, 6C and 6D, which show the image processor 20 at different stages in the four-step iteration. The several FIGURE 6 drawings are thus identical except that each shows dif-ferent specific connections which the program control unit 56 provides between the elements of the processor. Further, each FIGURE 6 illustration shows wi-th a heavy line the path of infor-mation transfer between the processor el.ements for a specific step in the operating sequence.

!.` "

3~

-3g~
~ he FIGURE 6 drawings also ~esignate the functions which each illustrated look up table 32, 40, 46 and 54 can provide, and FIGURE 5 shows a graphieal re~resentation of each function. With particular 5 reference to FIGURE 6A, the look up table 32 can provide either no transormation or the negative function of F`IGU~E 5A. The look up table 40 in the second refresh channel can provide either no transformation, an expand to nine-bit function of FIGURE 5C, or a compress to lO seven-bi~ function of FIGURE 5B. The feedback look up table 54 likewise can provide any of three unctions, i.e. no transformation, a compress to eight-bit f~nction of FIGURE 5D, or a linear and reset function of FIGURE 5E.

Each look up table 32, 40, 46 and 54 thus can assign new ~alues to each byte applied to it. The eight-bit bytes which the illustrated refresh memories 28, 36 and 44 store can have any of (256) posslble values. The FIGURE 5 functions hence show the new 20 values which the dirferent look up functions assign to each of the possible (256) values, and to the possible values which the sum of two bytes can have.

FIGURES 4 and 6A show that the aforementioned 25 first-step operation of pairing pixels is carried out by scrolling the contents of the first channel refresh memory 28 ~ith the scroll device 30. The scroll, which can be of any magnitude in either or both directions, is designated as being of (xn, Yn) pixels. The subscript 30 (n) identifies the number of the iteration being performed, inasmuch as different iterations in an operating cycle can -- and generally do -- involve _ao different scroll steps. The first illustrated step of the eighteen-itera~ion operating cycle which FIGURE 2 in part shows involves, by way of example, a scroll of (12~, 0) at this juncture. ~ere pertinent, FI,URES 4 5 and 6 show the block representation of each refresh memory with a coordinate desiynation in ~.he upper right corner to indicate whether the contents correspond to the origin pixel or to the comparison pixel of a pairing. FIGURE 6A accordin~ly shows the refresh memory 10 28 with the origin-desi~nating coordinates (o,o).

FIGURE 6A also shows tha~ the data path which the program control unit 56 (FIGURE 3) establishes in the processor 20 for this scroll opera.ion applies the 15 contents of the first channel refresh memory 28 to the scroll device 30, which introduces the specified scrollO
The scrolled image information is applied to the look up tab].e 32, which transfers it without change to the adder 48. The adder has no other active inputs and hence 20 simply applies the same scrolled image information to the feedback look up table 54. This Plement applie~ t'ne linear and reset function of PIGURE 5E, which in this instance impar.s no transformation to the data, so that the scrolled image information output from the scroll 25 device 30 is applied to the refrPsh memory 36 of the second channel 34. The memory stores this information with an operation subsequent to FIGURE 6A.

FIGURE 4 designates the foregoing scroll and 30 store operations in the first portion of Step I. At this juncture, the radiance information for each pixel of the original image is paired with the radiance information of the pixel offset from it b~ xn and Yn pixels~

3~

-41~
The illustrated im"ge processor 20 executes the remaining first-step operation of computing the ratios of paire~ radiance values hy applying both the negative conten~s of the first channel memory 2~ ancl the contents of the second channel memory 36 ~o the adder 48, and by storing the resultant sum from the adder in the memor~ 36. FIGURE 4 designates these operations in the last portion of Step I. FIGURE 6B shows that to provide this operation the program control uni. 56 10 applies the contents of the first channel memory 2~
through the scroll device 30, which imparts no scroll, and actuates the look up table 32 to provide the negative function and apply the transformed data to the adder 48. The control unit also applies the scrolled lj image information in the second channel memory 36 through the scrolled device 38, again without scroll, and through the look up table 40 ~ithout transformation to a second input of the adder 48.

The adder sums the two binary input signals and applies the resul~ant nine-bit byte to the feedback look up table $4, which the control unit 56 actuates to provide the compress to eight-bit function (FIGURE 5D).
This ~unction changes the nine-bit byte output from the 25 adder 48 to an eight-~it representation. The purpose is merely to accommodate a memory 36 which has only an eight-bit byte capacity. t~ote that each FIGURE 6 drawing shows the contents of the refresh memories 28 and 36 and 44 at the beginning o, the operation which it 30 depicts. Thus, FIGURE 6A shows the contents of the refresh memories at the beginning of the scroll operation, and FIGUR~ 6B shows the memory contents at the beginning of the ratio-computing operation.) To execute the second step of an iteLation, which forms the product of -the ra-tio with a quantity designatecl all old produc-t and which may be existiny or assigned, the progralrl con-trol unit 56 conditions the channel two and channel three elements to sum the cont~-nts of the refresh memories 36 and 44, as F:LGURE 6C shows.
The resultant intermediate produc-t is stored in the refresh memory 36/ all as the Elow chart oE FIGURE 4 designates for S-tep II ard in accordance with Equation (2) More particularly, with reference to FIGURE 6C, the ratio information in the refresh memory 36 is applied through the scroll device 38, without ofset, to the expand function of the look up table 40. This transformation expands the eisht-bit representation of each number to a nine-bit byte, FIGURE 5C. This expand operation restores to the information read from the refresh memory 36 the nine-bit format it had prior to the compress to eight-bit function which the feedback look up table 54 imposed during the prior, ratio computing, step.
The nine-bit output from the look up table 40 is applied to the adder 48, as are the contents of the third channel refresh memory 44. For the first iteration of operation, this memory stores an initialized old product for each pixel, preferably correspon-ding -to a unlform field of 100% ~eflectance as pre~iously discus-sed. The adder 48 sums the two inputs to develop the intermed-.ate product.
The feedback look up table 54 performs the Step III reset operation on the in-termediate product from the adder 48 as FIGURE
6C shows. The reset intermediate product is applied to the second channel refresh memory 36. Each byte input to the tahle 54 is V~38 a sum from the adder 48, and the reset portion of the FIGURE 5E
function is effective at this juncture. This function transforms input numbers valued between (-255) and (0) ~o (()), transforms input numbers with a ~alue between (0) and ( 255) to an eight-bi-t number oE identical value, and transforms input nl,mbers with a value greater than ( 25~) to the maximum value of ( 255). This reset func-tion thus produces a field of eight-bit reset products which are limited to values be-tween (0) and (255), and in which inpu-t values both lower and greater than this range ave effec-tively clipped. Prior to this reset operation, each intermediate product identifies at least in part a lightness value. The reset operation normalizes the values of the highest lightnesses in the image being produced, regardless of the radiance detected or sensed from tha1 pixel in the original scene.
The reset intermediate products output f~om the feedback look up table 54 are stored in the seconct channel refresh memory 36 at the end of the Step III operation, as appears in FIGURE
6D, which shows the condition of the memory 36 prior to the next operation~
This next operation is the iteration Step IV determina-tion of a new average product ~or each pixel. The two sets of logarithmic numbers to be combined for forming the new product are in the second channel refresh memory 36 and in the third chan-nel refresh memory 4~. However, -the set of numbers in the former memory is scrolled by (x) and (y! units from the initial coordin-ates, as a result of the scroll operation performed in Step I, -43a-wher~-as the contents of the latter memory are ~:t the ini-tial co-ordinates designated (o, o). The reset l)roduc~t con-tents of the refresh memory 36 3~
--Aa~--accordingly are scrolled bac~ to the initial coordinates wi~h the scrol] device 38 by the requisite (~xn) and (~Yn) coordinates, as FIGURE 6D shows, before being applied to the adder 484 With further reference to FIGURE 6D, the second channel l~o~ up table 4~ compresses the reset products to seven-bit bytes with the function of FIGURE
SB, ~he thlrd channel look ~p table 46 imparts a similar compress function to the old product contents of 10 the refresh memory 44. The resultant two sets of numbers are applied to the adder 48. The compress operations imposed by the look up tables 40 and 46 yield from the adder 48 an eight-bit sum which can be stored directly within ~he eight-bit capacity of the refresh 15 memory 44.

More importantly, by compressing each set of numbers from the memories 36 and 44 in this manner prior to ~he addition, the resultant sum from the adder 20 represents an equal weighted, i.e. fifty-fifty, average of the two sets of numbers. Where unequal wei~htin~s are desired for the averaging, the look up ~ables 40 and 46 can ha~e appropriate compressing functions to yield a sum which is the appropriately weighted average. A
25 further reason for the compress to seven-bit operation prior to addition is described hereinafter.

FIGURE 6~ also shows that the feedback look up t ble 54 applies the linear and reset function (FIGUP~E
30 5E) to the resultant sum from the adder 48~ and that the output from the table is applied to the ~hird channel refresh memory 44. The numbers input to the look up 3~

.able 54 are all within the (0) to (255) range, and accordlngly the table applies the linear portion of its function to the sum, i.e., applies the sum from the adder to the memory 44 without transormation.

FIG~RE 4 shows the resultant averaged new product in the refresh memory 44 at the completion of this Step IV operation in terms of equation (4).

The FIG~RE 3 image processor 20 is now ready lO to repeat the four-step iteration~ Each iteration after the first one uses the same original image field of radi~nce information which the first channel refresh memory 28 initially stored. Howe~er, the succeeding iterations do not use the initialized conditions in the 15 thi-d channel refresh memory 44, but rather use the new averaged product as computed and stored in that memory in the last step of the preceding iteration.

The FIGURE 3 system performs a number of the 20 iterations described with reference to FIGURES 3-6 an~
compares different pixels in each iteration. The comparisons are made by scrolling the contents of the refresh memories 28 and 36 with the scroll devices 30 and 38 through different coordinate distances in each 25 iteration. ~he theoretical objective is to comp2re the original image radiance information, stored in the first channel memory 28, for each pixel with that stored for every other pixel. FIGURE 2 shows a su~-set of twelve scroll displacements. It is a measure of the 30 significance of the invention that a complete image-producing cycle requires only eighteen such itera.ions.
The operation commences with a largest displacement, and 3~3 -4~-proceeds ~hrough the cycle with successive sets of two iterations, each of which involves progressively smaller scroll displacements. Note that in every iteration except the final two, (i.e. numbers seventeen and S eighteen), each pixel is paired for comparison with a pixel spaced from it by more than one pixel unit. That is, all but the final two pairings are between pixels that are separated with at least one pixel be~ween them.
The array of averaged new products available from each 10 processor 20, 22 and 24 at the completion of these eighteen iterations is applied, under control of the program control unit 56, to the ou~put stages 58, 60 and 62. ~his completes one operating cycle of the ~IGVR~ 3 system.

15 Threshold With further reference to ~IGURE ~, where it is desired to impose a threshold on the ratio of radiances determined in each processing iteration, the feedback look up table 54 can provide this operation.
20 The inset in FIGURE 5D shows a compress to eight-bit function which also provides a threshold function. This illustrated threshold function is such that all input values between (-2) units and (~2) units inclusive produce the same output value. Such a threshold 25 function can be advantageous in an image processor accordins to the invention to yield a ratio of unity when the two radiance values being compared are within a specified percent of each other. This threshold removes the spatially slow changing effects or gradients of 30 illurnination found in many images. As desired, a suitable threshold does not visibly affect the accurate imaging of discontinuous changes in radiance ,.., . ~, .",~

inrormation. The latter radlance changes or transitions, ~hich are the ones measured with the practice of this invention, generally stem from the changes in reflectance that occur at the boundaries or 5 edge of an o~ject in the original scene or ima~e. ~y way of specific example, \~here the memory levels (O) to ~255) evenly represent four log units of radiance, the foregoing threshold suita~ly treats radiance values that are within seven percent (7%) of each other as being 10 equal. This value is not critical; other values can ~e used as appropriate for the implementation and the application.

The threshold operation is applied to the 15 ratio ou~put from the adder 48 in Step I of the iteration described above. That is, the feedback look up table 54 imposes the threshold 'unction simultaneous with the compress to eight-bit operation discussed with reference to FIGURE 6B. It will be appreciated that a 2~ threshold is to ~ome extent inherent as a result of quantifying data in the present embodiment of the invention. The practice of the invention as described with the system of FIGURE 3 hence can be considered as imposing a threshold on each ratio computation.

A further finding of this invention is that a threshold alone is insufficient to remove from all images the undesirable effects on total dynamic range of gradual changes in illumination. When a uniformly illuminated image has pixel-to~pixel variations in 30 radiance which regularly exceed the threshold, the threshold alone is not sufficient to remove gradients superimposed on that image. The problem is typically ~ 3 8 encountered in images with s gnificant pixel-to-pixel sisnal fluctuations which are introduced by limitations of the image detecting mechanism. Within a single image object, these variations commonly exceed the thresholds 5 of plus or minus one or two grey levels which typically are adequate in fl~ctuation-free images. This degree of consistency exceeds the tolerance levels of typical electronic image systems. Even with low-fluctuation images, a oredominance of minute object detail can have 10 similar effects. Despite the already demonstrated abilities of a threshold process, other techniques are important for many images to reduce the influence of gradients on the dynamic range o~ calculated liahtness fields.

One mechanism alterna,ive to a thres~301d for gradient removal is a system that combines a reset step with a lightness field-determining operation in which many different comparison segmental areas influence each segmental area. Each segmental area has a dir~ferent 20 history of spatial interactions with other areas. When the history of interactions is limited, the influence of random fluctuations is propagated along these limited directions and causes local areas of unwarranted higher or lower ~ightness. This unwanted propagation of random 25 fluctuations does not occur when each segmental area is influenced by very large numbers of comparison segmental areas. Instead, the random events cancel one another, In addition, a gradient is by definition a 30 radiance change in a particular direction. The combination measure -which this invention provides does not emphasize radiance gradients. The contributions of ':.

3~
..

such gradients are different in magni-ude for each direction and spa~ial parameter of comparison. Further, a gradient produces a smaller change in lightness field calcultions than an object edge of the same magn}t~de.
5 An extended edge is equally detectable in all directions that cross it. It is considered that the combination measure may emphasize radiance changes produced by abrupt edges because most di~;ections and most spatial parameters for grouping segmental areas for comparison lO yield the same measure of the change. ~ence multiple measuring iterations as described herein yield measures which reinfor~e one another.

In the study of human vision one often finds 15 in the literature the division of segments of ~isual images into two arbitrary categories: objects, and illumination~ In a~dition, the literature contains numerous discussions of how human vision discounts illumination, so that information about objects in the ield or view has greater emphasis. This arbitrary division of visual images has many exceptions. For example, shadows produce large changes in sensation, despite the fact that they are intensity variations in illu~ination. As another example, gradual reflectance 2~ changes across the surface of an object cause small changes in sensation, despite the fact that they represent changes in the properties of the object.

Instead of characterizing different portions 30 of images as objects and as illumination, it is more useful to characterize them as radiance transitions that are abrupt, or as radiance changes that are gradual.
Radiance transitions that are abrupt generate large .~

3~
~ Q-changes in ligh.ness, ~hereas radiance transi~ions that are gradual generate small changes in lightne~ss. Siynal processing systems tha~ produce lightness fields produce quantities tllat correspond to lightness.

The foregoing lightness irna~e processing of this invention realiæes these properties of visual processing by calculating combination measures in such a way that abrupt change_ in radiance are characteri~ed by a set of reports all of which are the same.
10 Furthermore, combination measures are calcula~ed in such a way that gradients are de-emphasized by either a threshold or a technique using many comparison segmental areas with diffe.ent spa~ial interaction histories, or both.

15 Wra~around Insulation Each image processor ~0, 22 and 24 of FIGURE 3 provides the foregoing measure of a radiance transition n a manner different rom that previously described in -he case of a pixel that is paired with a location that i~s beyond a boundary of the image field. '~is 'i_ferent operation, termed wraparound insulation, minimizes errors that otherwise can arise from the foreooing determillation of a radiance transitior.
measure, e.g. a new product, for a pixel located such 2- tnat after the 5croll displacement it is to be compared with an out-of-field location, i.e. a location that lies beyond a boundary of the image field being processed.

The image processor which the invention 30 t ro~-ices avo:ds ~his rror by identifying, in each . eration, ~ch pixel 'hat is to be compared with an ~ ~, ., ~ J

out-of-field location. m e ~rocessor retains the old product for that pixel, and uses it as the new product.
Thl~s feature of the invention i5 described with reference to an illustrative iteration diagrammecl in 5 FIG~RES 7A, 7B, 7C and 7D for the image processor 20 and which involves pairing each pixel with the one located (12~) pixel units to the left, Each FIG~P~E 7 drawing shows only a portion of the processor 20, and each designates the contents of each illustrated refresh 10 memory as mapped in four equal sized regions, each of (128) pixels by (~12) pixels, The original image information which ~emory 28 stores in the memory regions 2~a, 28b, 28c and 28d is designated as A, B, C and D, respectively~

The illustrated image processor 20 handles the out~oF-field situation which the scroll of (12~) units presents in a conventional manner. As FIGURE 7A shows, the processor scrolls the information A, which is originally stored in the leftmost region 28a of the 20 first channel memory, to the opposite side and places it into the right edge region 36d of the memory 36 in the second channel.

FIG~RE 7A, which thus corresponds to FIGURE
25 6A, E~rther illustrates this wraparound scroll operation with the representation that the contents of the refresh memory 2B, upon being scrolled (128) pixel units to the left, have the format which appears in the memory map 64 shown to the right of the adder 48. This mapping of 30 information is stored in the second channel refresh memory 36 by way of ~he feedback path 52. FIGURE 7~
~nus shows the contents of the two refresh memories 28 3~

and 36 a ter the Step I scro'l operation. The shading designates the memory regions 28a and 36d which store information that the scroll operation wraps around from one memory border to the other, i.et the memory region 5 which in this exarnple stores the information A. Thus, in every iteration described above with reerence to FIGVRES 4 and 6, the image processor 20 scrolls the contents o memory elements adjacent an edge or boundary of the memory and wraps it around for storage in memory lO elements adjacent the opposite edge or boundary.

The transition-measuring image processing described above with reference to ~IG1~RE 6, however, can encounter problems if it processes the wraparound 15 information in the same manner as other information~
That is, imaging errors are likely to arise in khe present example if the radiance information A wrapped around for storage in region 36d OL memory 36 is processed in the same manner 25 the information in the 20 other regions 36a, 36b and 36c of that refresh memory.
An example of this error occurs where the original image being processed is of a scene illuminated from the right side with ten times more light than on the left side.
The lO-to-l gradient in illumination improperly 25 dominates the Step I ratio calculations where radiance information from pixels at opposite sides of the image field are compared by virtue of the scroll wraparound.

The processor 20 which the invention provides 30 solves this problem by disregarding transition measures which result from such wraparound ratios. The processor instead identifies each pixel where the radiance inormation is compared with information that is scrolled Lrom edge to edge, i.e. where a wraparound ratio is lnvolved. The processor retains the old product for each such identi~ied pixel and uses that product in the next iteration, instead of determining a 5 new product, as occurs for all other pixels. The processor 20, typical of the processors 2~ and 24, thereb~ insulates each iteration from communication between image locations that are separated in a direction opposite to tlle direction of the scroll for 10 that iteration.

FIGURE 7B shows the same portions of the processor 20, with the mapped contents of memories 28 and 36, as FIGUR~ 7A. The mapping 66 at the right of 15 FIGURE 7B depicts the logarithm of a ratio and hence depicts the sum of the negative of the memory 28 contents and the memory 36 contents output from the adder 48 after the Step X ratio computation o Equation (1). The riyht most mapping region 66d contains the sum 20 designated (A-D), which i5 a wraparound ratio, i.e. a diference between memory contents which in the first channel memory 28 are separated in the direction opposite to the direction of the scroll ~hat preceded the ratio computationO This sum manifests the 25 wraparound error discussed above. The contents of the remaining mapping regions 66a, 66h and 66c are correct and unaffected by the scroll wraparound.

The illustrated processor proceeds in the same 30 manner described with reference to FIGURE 6C with the Step II computation of an intermediate product~ FIGUR~
7C shows the mapping of the refresh memory contents for the second and third channels of the processor ~0. The 3~

contents of the memory 36 ha~e the same mapping as appears in the mapping 66 a~ the ~ight side of FIG~RE
7B. The reresh memory 44 contains old product information designated as PA, PB, PC, and P~ for the 5 four memory regions 44a, 44b, 44c and 44d, respectively.
The mapping 68 of the sum of these two memory contents witn the adder 48 appears at the right side of FIG~1RE
7C~ This sum is the intermediate product cvmputed according to Eq~ation ~2). The sums mapped in regions lO 6~a, 68b and 68c are in the desired form, unaffec~ed by the scroll wraparound. The sum mapped in region 6~d, ho~ever, is prone to scroll wraparound error.

The processor operation proceeds to the Step 15 III reset operation ~hich the feedback looX up table 54 performs, as described above with reference to FIG~RE
6D~ The reset intermediate product is stored in the second channel refresh memory 36, a~ FIGURE 7D shows.

To execute the last iteration step, i.e. the Step IV averaging computation, the scroll de~ice 38 scrolls the contents of the second channel memory 36 by an equal and opposite amount from the scroll effected in the first step of FIGURE 7A, i.e. a scroll of (128) 25 pixels to the right in this example. The mapping 70 in FIG-~RE 7D represents the second channel refresh memory 36 contents after this scroll operation. The field of information is applied to the adder ~8 by way of the look up table 40. At the same time, the contents of the 30 channel three memory 44, shown in mapping 72 juxtaposed with mapping 70, are applied to the adder 48 by way of the look up table 46.

~ he i'lustrated processor 20 combines the two sets or fields of informatlon in a straightforward manne~ for all regions of the mappings except for the reyion that includes information manifesting scroll 5 wraparound. This is the information in the channel two memory region 36d, and which th~ scroll device 38 scrolled to t~e mapping region 70a.

The processor 20 ~evelops the new product 10 information or this region in response entirely to the old product information in the mapping region 72a, i.e.
in the region 44a of the third channel memory 44. As FIG~RE 7D shows, the resultant mapping 74 of the adder ~8 output contains old product information in the region 15 74a. This mapping region corresponds identically to the ~emory 28 resion 28a designated in FIGURE 7A as containins information which this iteration would subject to a scroll wraparound. The remaining mapping regions 74b, 74c and 74d contain new products computed 20 as described above according to Equation (4). The modified new product which the mapping 74 represents is stored in the channel three refresh memory 44, in accordance with the flow chart of FIGURE 4, Step IV.

~5 One ~etailed operating sequence for determining a new product which retains old product information in the foregoing manner, as mapped in the memory region 74a, is to save the old product inormation in a buffer or other store of the program 30 control unit 56, ~IGURE 3, prior to the product-producing addition operation of Step IV. A subsequen~
operation writes the saved old product information into the specified region of the third channel refresh memory 3~

44. The program control unit 56 can identify the re~ion of memory 44 which contains old product information to be saved in this manner ~y using the coordinate dicplacement information, i~e. (xnl Yn)~ which controls 5 the scroll devices 30 and 38 for the iteration in process.

Anotller operating sequence for effecting the foregoing wraparound insulation involves inserting a 10 marker digi~ in each refresh memory location which stores information that is to be subjected to a scroll wrap2round. The look up tables 40 and 46 can effect a compress function, with the sets of marked infcrmation, different from that of FIGURE 5B to yi eld from the adder 15 48 an averaging in which the resultant for the marked memory locations is responsive exclusively to the old product information. More particularly, according to tnis alternative sequence, after the reset intermediate product is stored in the second channel memory 36 to 20 complete Step III (FIGURES 3 and 7D), the program control unit 5~ clears the low order bit of every byte of product information in both refresh memories 36 and 44. The control unit next writes a binary ONE into the low order bit o only those bytes which the register of 25 the scroll device 38 identifies as being lnvolved in a scroll wraparound for that iteration. This selective storing of binary ONES tags or marks each byte of product information in the mem~ries 36 and 44 to identify those which reflect scroll wraparound. The 30 marked contents of the memory 36 are next scrolled in the usual manner with the scroll device 38.

,However, instead o using the equal-weighting compress function of FIGURE 5B as discussed above with 3'V~3,~

reference ~o FIG~RE 6D, the second channel look up table 40 provides a selective-averaging transfer function shown in FIGVRE 8A, and the third channel look up table 46 provides a selective averaging transfer function 5 shown in ~IGURE 8~. FIGURES 8A and 8B show each transfer function for in~uts of decimal magnltude (0) to (10): each function ranges in the same manner shown for inputs of magnitude (0) to (255).

Each selective-averaging transfer function processes a byte of marked information differently from a byte of unmarked information. More particularly, each marked byte is an odd-valued number because it has a binary ONE in the lowest order bit plane, whereas every 1~ unmarked byte has a binary ZERO in the low order bit and hence is valued as an even number. FIGURES 8A and ~B
designate the transfer function for each odd-valued input with a cross, and a circle designates the function for each even-valued input.

The FIGURE 8A transfer function produces a zerovalued output in response to every odd-valued input, and produces a seven-bit output valued at one-half the magnitude of each even-valued input. That is, an input byte of decimal value 1, 3, ~, 7...255 produces an 25 output byte of value zero, whereas an input byte of decimal value (8), for example, produces an output value of decimal (4). With this transfer function of FIG~RE
8A, the look-up table 40 in the second refresh channel applies a zero-valued input to the adder 48 in response 30 to each marked byte, which is a byte involved in a scroll wraparound, and applies a one-half value byte to the adder in response to every unmarked input byte.

3~

-5~-The look up ,able of FIGURE 8B likewise p~oduces a one-half valuecl output for every even valued inpl1t byte. ~owever, it responds to each odd-valued input byte to produce an ou~ut value o' the same S ma~nitude. ~ith this transfer function, the look up table 46 responds to each unmarked input byte to apply a byte of one--half the input value to the adder 4~, but responds to each marked input byte to apply a byte of the same value to the adder 48.
The adder 48 responds to the fields of intermediate product and ol~ products transformed with these 'unctions to produce the desired "insulated" field of new products which FIGVRE 7D shows with the mapping 15 74.

A Two Channel Image Processor 3efore considering a further embodiment of the invention, note that the image proceseor 20 described 20 with reference ~o FIGUR~S 6 and 7 employs three refresh channels. The first stores original radiance information regarding the image being processed. The second channel is used to perform calculations in accordance with the lightness imaging process summarized 25 in the flow c`hart of FIGURE 4. ~he third channel stores new product information and presents it as old pro~uct information for the next iterationr In cQntrast to these features, FIGURE 9 shows another image processor 80 in accordance with the invention. ~he processor 80 30 attains the same results as the processor 20, but with two memory channels, Further, it uses delay lines to effect scrolling in a time domain, in contrast to scrolling in a spatial domain as in the processor 20.

More partictllarly, the signal processor 80 has a first refresh channel 82 with a refresh memory 84 ~hat receives information from an input stage 86. l'hc contents of the memory 84 can be applied to different 5 inputs o an array adder 88 by way of ei~her a neyative look up table 90 or a delay line 92. A second refresh channel 94 of the processor 80 has a memory 96 connected to apply the con~ents to a furt~er input of the adder 88 or to a second adder 100 by way of a further delay line 10 g8. A reset look up table 10? is connected to receive the output from the first adder 88 and apply it to the second adder 100. The output from the adder 1~0 is applied to a lo~ up table 104 having a compress to eight-bit function as shown in ~IGUP~E 5D. ~he output or 1~ the look up table 104 is the output from the processor 80. A feedback path 106 applies this output to the input of the second channel memory 94. The processor 80 operates in conjunction with a program control unit 108.

~0 The operation of the PIGURE ~ processor B0 typically commences in the same manner as described for the processor 20 with the first channel memory 84 storing a field of original image information proportional to .he logarithm of the radiance at each 25 element of the image field. The second channel memory 96 is initialized with the logarithm of a selected uniform radiance ield. ~he processor 80 reads information from the refresh memories on a time sequential basis characteristic of conventional shift 30 register systems. With such operation~ time after onset of a memory scan or read operation is directly related to pixel location in the image field being processed.
That is, the information for successive pixels is read

-6~-ou~ at known success ve times in a memory read cpera~ion. Further, each ~emory 84 and 96 can advantageously use two orthogonal interconnections of the memory elemenks therein, in order to accomplish 5 image displac~ments in either the x (horizontal) direction or the y (vertical) direction with less delay than the time required to read a single line of information from either memory.
. .
The processor 80 simultaneously performs the calcuiations of Equations (1) and (2) to produce an intermediate product. For this operation the processor simultaneously sums, with the adder 88, .he three sets of information identified by the three terms on the 15 right side of Equa~ion (2), i.e. log op(o,o), log r(x,y) and the negative o~ log r(o,o~. To effe~t this operation, the contents of the memory 84 as made negative with look up table 90 are applied to one input of the adder 88, the contents of the memory 96 are 20 applied to another input of the adder, and the contents of the memory 84 are applied to a third input of the adder by way of the delay line 92. The delay line introduces a time delay that displaces the information being read sequentially from the memory 84 in time by an 25 amount equal to the desired (x,y) scroll. The processor 80 thus generates the intermediate product at the output from the first channel adder 880 The intermediate product is reset with a 30 transformation function similar to that oF FIGVRr 5E by applying the signals ou.put from the adder 88 to the look up table 1020 For memories 84 and 96 that store radiance information as eisht-bit bytes, the reset function for table 102 transrorms ten-bit bytes, which the adder ~8 produces in summing ~hree eight-bit input bytes, to eight-bit bytes. I~e reset produc~ is applied to the second charlnel adder 100, which s~ms it with S il~formation identified by the additional term on the right side of Equation (4), i.e. the loy op(x,y) term.
This infor~a~ion is in the memory 96 and is applied to the adder 100 by way of the delay line 98, which imparts the same time delay as delay line 92. Th~ look ~3 table 10 104 compresses the resultant summation si(nal output from the adder 1~0 to effect the divide b two operation for completing the ave-aging computation of Equation (4~. The resultant averaged product information is applied by way oS the Feedback path 106 to the second 15 channel memory 96, where it is written on the same time sequential basis ~ith which the memories are read.

The FIGURE 9 processor 80 further has a buffer 110 to s,ore the information in the memory 96 that is 20 combined with wraparound information from memory 84.
The program control unit 10~ reads the buffer-stored information bacX into the memory 96 in place of the new product in~ormation which results from locations that are paired by way of a time-delay scan wraparound~ This 25 use of the buffer 110 provides wraparound insulation in the manner described above with reference to PIGURES 7A
through 7D.

Three image processors identical to the 30 processor 80 of FIGURE 9 can provide a full-color image producing system liXe the system of FIGURE 3. That i 5, ~he system of FIGURE 3 can be constructed with a two-channel processor 80 (FIGURE 9) in place of each .. ..

processor 20, 22 and 24. It will also be apparent that the inventlon can provide black-and-whi~e ancl othcr single color imaging with a svstern as shown in FIGURE 3 which has only a single image processor, instead of 5 three as shown. Further, a single-processor sys~em can calculate all three lightness fields for full color imayiny with a time-sharing operation, i.e. by processing the red, the green, and the blue components separately on a time sequential basis~

lO Multi~Size Pairings -The embodiments o~ the invention described above with reference to r IGURES 3 through 9 employ pixels which represent imaae sections of uniform size.
Also, the spacings between elements paired in each 15 iteration with these embodiments bridge a variety of distances and extend in different directions. The invention can also be practiced using diferent elements to represent image sections of different sizes. As wi~h the foregoing embodiments, these further embodiments can 20 be practiced using independent iterations or averaged iterations, i.e. using either a reset interme~iate pro~uct or an avera~ed product as the final combined measure for each iteration.

A first embodiment of this practice of the invention with image elements of different sizes is practiced with the system of FIGURE 3 using the image processors 20, 22 and 24~ although it can equally be practiced with the same system using the processor 80 of 30 FIGURE 9 in place o each processor 20, 22 and 24. As ~ill become apparent, the embodiment employs refresh memories with greater storage capacity than in prior embodimen's ~o attain the saMe level of image resolution. In this practice of the invention, the radiance information identifying the image to be processed is recorded or otherwl~e stored, as in a 5 menlory element of the program contxol unit 56. q~e program control unit 56 also recorcls the orlginal image information, as thus recorded or stored, in a portion 120 of the first chanllel refresh memory 28, as FIGURE 10 shows. The identica] image information is also copied ~0 into four other portions of the memory 28 with different reduced si~es and orientations, as mapped in FIGURE 10 in the memory portions 122, 124, 126 and 128. The illustrated mapping thus stores five image representations, each perpendicular to and one-half as 15 large as t~e next larger one. Purther, the several image representations preferably are stored spaced apart.

Those skilled in the art will recognize that 20 other devices can be used to generate different size images of the same scene on a single detector array.
Except for tlle different orientations, the same image representations as shown in FIGURE 10 can be produced by using five lenses of different focal lengths and five 25 different lens-to-detector distances. The shortest focal length lens forms an image representation analogous to the one mapped in the memory portion 128 in FIGURE 10. Each successively larger image representation is made with a longer focal length lens 30 that makes the image representation twice the size of the previous lens. The l~ngest focal length lens forms the largest image representation, analogous to the one mapped in the FIGI~RE 10 memory portion 120. ~nis ~v~

embodiment does not require a high resolution record of the entire image in the program control unit 56 because eacl~ lens in the system makes individual minified image representations of the scene. The imagè
S representations, whether optically or electronically minified, are processed in the same manner.

The FIGURE 3 system computes transition measures for each image representation, in the same 10 manner as described above with reference to FIGURES 6A
through 6~ and the flow chart of FIGURE 4 (and with wraparound insulation according to PI&T~RES 7A-7D), to compute a field of new products for each of the multiple image representations mapped in the memory 28 as shown 15 in FIGURE 10. ~he scroll in each iteration, however, is selected to compare only elements of the same image representation. That is, the image representation stcred in each memory portion 120, 122, 124, 126 and 128 is compared only with elements within that memory 20 portion. Comparisons between different image representations are avoided or are processed as scroll wraparounds in the manner described above with reference to FIG~RE 7. Wi.h perpendicular or like differently-oriented image representations as in FIGU~E 10, the 25 system operates with uniformly-directed scrolls, w~ich can yield economies in implementation~

~ pon completion of the selected num~er of such iterations, the final new products for the several image 30 representations in the third channel refresh memory 44 represent the results of five independent lightness imaging calculations that use diverse path directions to compare elemental areas of different real sizeO 1'he V~3~3 program control unit 56 electronically zooms and rotates the resultant average products for the differen~ image representa~ions to identical size and orientation, and t~en averages the several new product fiel~s to form the 5 ~inal single field of lightness-imaging products.

Table I lists, in order, the scroll coordinates for a two-hun~red~iteration operating cycle for processing different image representations as shown 10 in ~IGURE 10 in the Coregoing manner. The Table identifies each iteration by numbert i.e. (1) through (200~, and presents the (x) and ~y) scroll displacement coordinates for each iteration. The practice of the invention as described with reference to FIGURES 3 and lS 10 and Table I employs larger capacity memories to store the multiple image representations with equivalent resolution, as compared to the practice according to Figures 3 and 6.

(The re~ainder of this page is intentionally blank,) ~abie 1

7 -1 1 66 0 -1

8 -1 0 67

9 -1 -1 68 -1 0

10 0 -1 69 -1 -1

11- -1 -1 70 -1 0

12 0 -1 71 -1

13 -1 -1 72 0

14 0 -1 73 -1 lS 1 -1 76 0 2~ -1 0 81 3i -1 -1 90 0 33 -1 -1 g2 0 34 0 -1 g3 -1 36 0 -1 gS -1 3& 0 - 1 97 h3 -1 -1 102 0 -47 -~ 1 106 0 -~9 -1 1 108 -1 0 56 -1 ~ 113 56 -1 0 llS

Sr30 -1 117 ~9 -1 -1 118 0 ~sble 1 - cont~nued 1211 1 lB0 0-1 ~2 1 0 161 1-1 1260 -1 lB3 1251 1 18' Q

1271 1 lBS 10 1291 1 lBB 10 1340 -1 1~3 1360 -1 l9S 1-1 13g1 1 196 -10 1~1O 1 0 199 -1 1~4.2 0 1 200 -1 0

15~--1 0 Another embodlment of the invention with pairings of multi-size image sections is identical to that described with reference to FIGURE 10 except that each image representation i5 processed with a sequence 5 of iterations ~lich may differ ~rorn the sequences used to process other image representations~ Each sequence has approximately the same num~er of iterations, but the magni~u~es of the distances between paired segmental areas ar~ tailored for each si7.e of image 10 representation. It will also be apparent that each image representation rnapped in FIGURE 10 can be processed with a separate image processor tailored for the size of whichever image representation it processes.

15 Se~uential Progression of Sizes Alternative to processing the different image representations independently of one another, as described with reference to FIGURE 10, improved image quality after equal or fewer iterations results when the 20 several image representations are processed sequentially and the new product determined for each image representation is used as the initialized information for the third channel refresh memory 44 (or for the second channel memory 96 when each processor is of the 25 two-channel construction shown in FIGURF 9). It is further advantageous to process the image representations in this c,rdered sequence starting with the smallest image representation as shown in FIGURE 10 in memory region 128, and progressing to the largest.

FIGURE 11 shows a full-color image-producing system for this practice of this invention. The illustrated system has an input stage 130 ~hich applies ~ 8~?~38 image-responsive radiance information for each of three color bands -to difEe.rent ones of three memories 13~, 134 and 136. A
prograTn contxol ullit 1.38 i.llustrated as con.nected wi-th all other elements of t.he system beyond the inpu-t stage, con~_rols the opera-tion of three imac3e processoxs .l40, 142, 144, illustrati.ve'y each of the two-channel construction described abo~e ~ith reference to FIGURE 9. Each processor 140, 142, 144 is connected -to receive image information from one memory 132, 134, 136, respectively, by way of a zoom stage 145. An output stage 146 ~eceives the proce-;sed image signals from the three processcrs to provide the color masking, exposure c:ontrol and like further signal processing as appropriate and to provide the desired display or other output presentation of the lightness processed image.
The system first stores the logarithm of the radiance information for each element of the image field, for each of three wavelength bands, in different ones of the memories 132, 134 and 136. The program control unit 138 reduces the size of the informa-tion field in each memory 132, 134 and 136 by a factor of sixty-four with the zoom stage 145, and stores the minified image field in a one-sixty-fourth portion of the first channel refresh memory 148 of each processor. The control unit 138 similarly initializes, preferably with the logarithm of a selected uniform ra(iiance field, a correspondingly located one-sixty-fourth portion of the second channel refresh memory 150 in each processor 140, 142, 144. The control unit then performs with each processor a selected number of iterations, each as described above with reEerence to FIGURE ~, using only the one-sixty-:Eourth portion of each memory which stores image information.

3~
~7~-Before the program control unit 138 executes a second cycle of iterations, the new product results of the last iteration of the prior, first cycle are ma~nified, with the zoom stage 145 and by enabliny data 5 paths 152 and 154 shown in the processor 144, by a linear factor of two to initiali~e the second channel memory 150 of each processor for the next set of lterations. The zoom stage 145 initializes each first channel memory 148 with a sixteen-to-one area reduction 10 of the information field in each associated memory 132, 134, 136, using one-sixteenth of the refresh memory.
The resultant new product in each second channel refresh memory 150 is again electronically magnified and placed in a one-quarter portion of that memory to initialiæe it 15 for the third cycle of iterations. Similarly, the first refresh memory in each processor receives, in a one-auarter portion, a four-to-one area reduction of the original image field information in its associated memory 132, 134 and 136. After executing the third set 20 of iterations, each processor is again initialized, this time using the entirety of each reresh memory. The results of the fourth set of such iterations yield in each second channel memory 150 a full-size field of lightness-identifying new product information for 25 producing the desired display of the image field.

The system of FIGURE 11 thus performs multiple cycles, each o~ multiple iterations and each using a successively larger ield of image-identifying 30 information. Each cycle produces a field of new product information which is electronically zoomed or magnified to form the initi~l old product for the start of the next set or iterations. ~I~URES 10 and 11 thus illustr~te practices of the invention using differently-siæed image elements in di~ferent processing iterations.

Table II is a list of thirty-two numbered pairs of relative time delays for one specific practice o~ the invelltion, used for each of the four cycles as described for the system of FIGURE 11.

TABLE II

2 1 1 1~ 1 1 4 1 1 20 ~ 1 15 5 1 1 21 ~ 1 6 1 1 ~2 8 1 0 24 ~1 1 ~01~ 1 0 26 -1 1 1~ 1 0 28 -1 1 13 1 -1 2g -1 1

16 1 -1 32 -1 Another embodiment of the multiple-size pairings uses a combination of optical and electronic 3~ techniques to change the size of the image. Here a zoom lens is employed to make a series of different size image representations of the same scene. The system fil-st sets the zoom lens to the smallest specifled 3~

representation o~ objects in the scene. The optical image under this condition includes more of the entire field of view because at this setting the lens has its widest angle. The desired portion o the scene is S ~maged in the center of the detector array an~
represents the image in relatively few pixels. The remainder of the image ormed by the lens is called the peripheral image. In some applications the proce~ss can ignore the entire peripheral image by using techniques lO analogous to those described regardiny wraparound insulation. In other applications, the process benefits from using the peripheral image information in calculating lightness fields that are influenced by segmental areas outside the final desired image.
In both cases, whether the perlpheral image is included or not, the lightness field is calculated in the manner described with FIGURE 9. The long distance, global interactions are performed first, in relatively 20 few iterations. The processed image is then zoomed by electronic means to twice its original size by rewriting each pixel in the preceding image representation as four pixels. If desired, any known shading element can be used to smooth the edges of each four-pixel area. The ~5 new enlarged image is sent to the product memory 96 in FIGUR~ 9 to serve as the array of previously determined values for the next stage of computation. The system next sets the zoom lens to form an image twice the size of the previous image and enters it at the input stage 30 86 in FIGURE 9. The process then computes the next stage of the calculation for the slightly less global interactions. The system repeats this sequence a number of times, and for each successive i~eration stage, the 3~3 contents of the second channel memory 96 are zoomed by electronic Means, and the image informa~ion in the input stage ~6 is zoomed by optical means, each time to produce images of the same size. The final calculated 5 iightness field has the benefit of ~oth global and local interactions. As descr;bed above, this system can be implemented so that the lightness field of the desired image benefits from the information in the ul~imately disc~rded përipheral images.

A Lightness Imaging Camera ~ IGURE 12 illustrates, in a schematic manner, an application of the invention to a self-developing camera 160. A light-tight camera housing 162 mounts a 15 lens 164 that focuses the desired viewing field onto the photosensitive surface of a multi element CCD array 166.
The array 166 includes the electronic circuits of a program control unit and of three image processors, together with color masking and exposure control 20 circuits as described above. A multiconductor cable 168 applies the resultant lightness imaging signals to a light emitting diode (LED) array 170. The lightness imaging signals from the CCD array 166 energize the arrayed light emittiny diodes to expose a film unit in a 25 film pack 172 through a transparent optical plate 174, which could, if desired, be lenticulated. The camera includes a pair of motor driven spreading rollers 176 and associated mechanism for withdrawing each film unit from the film pack after exposure and ejecting it from 30 the camera, as illustrated with the film unit 178, in a manner which initiates the self-developing process known for self-developing cameras such as those manuactured by the Polaroid Corporation.

This type of camera has a number of unique design fea~ures that take advantage of the physical properties of the components. ~he individual detector elements in the array 166 can be small, even in 5 compalison ~o tlle size of the individual elements in the light emittiny array 170. The latter array in turn is matched to the resolution of the fllm. The use of a small detecting array is advantageous because it allows the use OL relatively short focal length lenses for the 10 same size final image, i.e. film urlit 178, A short focal lenyth lens 164 means that the camera is proportionally less thick, and that the optical image has proportionally greater depth of field, thus making the requirements for focusing less demanding. In 15 addition, shorter focal length lenses have smaller f/
n~nbers for the same diameter aperture.

An array of charge coupled devices (a CCD
array) or OL charge induction devices (a CID array) is 20 particularly appropriate for the practice of this invention because o~ a combination of physical proper~ies, First, this type of p~otoelectric transducer is specifically designed to report the radiances o~ a large number of pixels~ Second, these 25 detectors respond to a large range of radiances. This enhances the ability of the system to correct for the large dynamic ranges of radiances ound in natural illuminations. Third, either array can additionally serve as a memory device. In principle this dual 30 property of the CCD or CID array, being both detector and memory, can lead to a multilayered embodiment of FIGURE 9 that includes at least an input stage 86 and memories ~4 arld 94 in a single integrated circuit chip.

-7~-Correspondingly, a FIG~RE 3 processor 20 can be constructed with at least the input stage detector 14 and the first channel memory 28 in a single IC chip.

S The embodiments described above thus provide eficient lightness~imaging systems that provide both local and global computational interactions of radiance information. A succession of the interac~ions compares radiance information from lmage locations that are 10 spaced apart by an ordered succession of distances, or compares radiance information of image representations having an ordered succession of magnifications. The embodiments further advance retinex processing by operation on a field basis, rather than on a location by 15 location basis, A practice of this feature with a field of (512)2 locati~ns, and employing two memories each having capacity to store the entire field, can rapidly calculate and store one-quarter million parallel one-step sequences of reset ratio products. The 20 advantage of this approach is that a single operation comparing two fields of information accomplishes - in this example ~ one-quarter million parallel computations. Succeeding iterations can build on the sequential product image in such a manner that N
25 iterations accumulate information at every location along a patterned N-jump excursion on an individual path. Compared with prior practices in which each operation deals with only a single pair of pixels, these eatures can achieve time savings over prior practices 30 of a factor approaching the number of locations.

The described embodiments manifest a further significant advance in processing ef~ic,ency by using field~by-field processing.

;r 1 ~ 3 ~3 Combining image-information fields produced with successive processing iterations in accordance with the invention has maximal productivity in conj~nction wi~h another eature which the foregoing embodiments S demonstrate, This feature is the pairing of image field ioca~ions employing a spatial parameter that changes from iteration to iteration. Every pairlng, and correspondinyly every iteration, brings to one pixel or location thé information accumulated by the pixel paired 10 therewith. This process yields a geometric growth in information accumulation such that there are (2N _ l) such accumulations after N iterations. One end result of a lightness-computing process which employs diverse interactions in this manner is a real time imaging 15 system.

Yet another feature of the invention which the foregoing embodiments implement is the ordering of iterations. ~le resetting function, which the 20 lightness~image processing of the invention employs to establish lighter areas as references, is a non-linear operation tha~ makes the orderi~g of iterations important. The combining of successive iterations in accordance with the invention with a geometric average 25 has been found to produce optimal results when the larger field displacements precede smaller ones. The resetting function preferably establishes reference to an e~treme or limit level of lightness, e.g. bright white or total blacX.

Three appendices are attached hereto.
Appendix I i5 a computer program listing for one specific practice of the invention as described above wi~ reerer,ce to FIGURES 3 and 6 and using either ~he eiahteen-s~ep cycle described with re~erence to FIGURE 2 or a fifty-six step cycle. T~e li~ting is in Fortran language and is prepared for use with an International 5 Imaging Systems ~odel 70 image processor controlled by a DEC PDP ll/60 computer operating uncler RSX llM.

Appendix II is a further description o a preferred form of the color mask stage 58 of YIGURE 3 10 for one prac~ice o the invention.

Appendix III further describes a preferred embodiment of the exposure control stage 60 of FIG~RE 3 for one practice o the invention.

It will thus be seen that the objects set forth above, among those made apparent from the preceding description, are efficiently attained. Since certain changes may be made in carrying out the above methods and in the constructions set forth without 20 departing from the scope of the invention, all matter contained in the above description or shown in the accornpanying drawings is to be interpreted as illustrative and not in a limiting sense. One change, among numerous ones, is that the invention can be 25 practiced with equipment which operates on an analog basis, rather than with the digital equipment of the described embodiments.

It is also to be understood that the following 30 claims are intended to cover all of the generic and specific features o the invention herein described, and all statemen~s of the scope of the invention which as a matter of languase might be said to fall therebetween.

3~

Having described t~e invention, what is claimed as new and secured by Letters Patent is set fo~th in the appended claims.

L3~
I-l APPEN_JX I
A program (named EYEF~Y) is provided with a data file (in lines 51-77 of the follo~ing program lis,ting), and the operator controls specific features of the sequence o iterations~ Two samples of this data file follow: The first specifies a set of eigllteen iterations as discussed with reference to Pigure 2. I~e second specifies a set of fifty-six iterations or use in producing a variety of irnages that exhibit the efSectiveness of the process.

~ ollowing the data files is a complete, commented, 277-line listing of EY~PLY, after which appear indices of program sections, statement functions, variables, arrays, labels, and functions and subroutines used. Finally, a separate list is included of the subroutines, with comments on the lunctions they serve~

Data File for 18--step Operation 18,0,0,1,0,2. 18 jwnps, controlled, (not used), nnulti-color,~eparations 0 to 2 BAZ4 ! zoomed face ima~e, scaled to4 decades o ! threshold-0 grey levels ~5,,, I weighting of old log(surn)=50%
256,0, ]

0,25~ ~

-128,0 ]

0,-12~ ]

64,0 0,64 ]

-32,0 ~ ! 18x,y pixel jump con~ponents, one pair/li~e 1~,0 0,16 -8,0 0,-8 ]
]

4,0 ]

0,4 ]

-2,~ ~

J~38 o,-2 ]

1,~ ]

O , 1 '~

3~

~ata File for 56-step Operation 56,0,0,1,0,2, 56 jumps, controlled, (not used), multi-color, separation~ 0 to 2 i3A~4 ! zoomed face image, scaled to4 decades O Ithreshold=Ogrey levels .5,,, Iwe.ighting of old log(sum)=50%
128,128 ]
-128,0 ]
-128,0 ]
0,-12~ ]
0,-128 ]
12~,0 ]
12~,0 ]
0,12~ ~
64~64 ]
-64,0 ]
0,-64 ]
0,-6~ ]
64,0 ]
64,0 0,6~ ]
32,32 -32,0 ]
-32,0 ~
0, 32 ]
0,-32 ]
32,0 32,0 ]
0,32 ]
16,16 -16,0 3~

- 1 6 , 0 ]
0,-16 3 1 56~c,y pixel jump components, one pair/line 0, -16 ]
16,0 ]
16,0 ]
0,1~ ]
~,8 --8,0 ]
-8,0 ]
0,--~ ]
0,--8 ]
8,0 ]
8,0 ]
0,-8 0,~-8 3 8,0 ]
8,0 0,8 4,4 ]
-4,0 -4,0 0,-4 0,-4 4,0 ]
4,0 0,4 ]
2,2 -2,0 -2, 0 ]
0,-~
0,--2 ]
2,0 2,0 3 0,~

~ ~3~

~-4 J~

F~-,R~ P_t'' ~--v2-51 20:31:22 22-JUL--90 P~ I
E ~.~ LY . .P;?~ R: ~ LI~WR
oor~ PRO:;RLV E~FLY
EY'PL`: o^es ins.3nt~neous Petinex prs_ossing Dn 3 (Ir~g-ener~y) im~e.
0'`^2 p'~LV-- n ~:,E-~ i5, V.XC.n~lOD
DD^' I~:r33rR IDI~X~0:7), IDIRY~0:7~ ,LLr;7~C,i256) ,LI';~-.I (2S6) ,~X.,l~l~' OD04 inte~er in~irx(O:7) ,ind~ 0:7) ,R~O~.V.,~f31se,fulimg ODD; inte~er jurn(rA~iter) ,nbs( r.xi~er~ ,iep(~ iter) ,n;:,:1(mxiter) ODD^ rr~al chi (rxiter) ,shi (r.~iter) .ehi (trxiter) ,Dh3(r,lxiter) 0007 r!~rE~,~R 3~:(~1~;' ) ,~cy~xcls~L) DODe teDI co(O:I) ooog integer buf f ~2048) DQ`D inee~et lutlin~255) ,~lP.avpDth~0:255,0:1) OOli ~ L~. ~ FC3(5),~,1C~(3),.scrach(0:255,D:I) OCi2 b~.e I~ORE,O!:,ISL~ P~l~:,Dn~rc,ever,n2Lne~3D),cor.. fil(3v),basls~30) DC13 by-e tnDrnt~3D) ,narnout(35) ODi4 ecr_'Ya~en e (fl~th(D,v),buff(l)),(lu~ne~ ,bu'f~102;)) OD'S r~.~ ;L~ ~0~ , 0,-1,-1,-1/
CrvlS ~1~ INDIR'' /1,1,0,-1,-1,--1, D, 1/
OD17 oldp,iho?,itt~t)~C~ D/ihop ~ (E'ihcp)/itot !Define csllbinino, fr3:tion C of 6uccessive iter~.jons ns e fur~ctjon oi bsp n~ber.
DCl~L` cell nstrn~l ~fcb,bL~,3,-S) C D~ LI Z E SY~.
C

C~
DOi9 Qll rbu~n(fc_,}cbu-.,icx,icy) D52D l:~D 3 IN=I, 255 OC2. rar~yrr~n ~ irl, ir2) D022 Lt~LTl.'.(N)tN-I
0023 Lr~ri~Ir(N)=2~(N-l28) !-~ill ex.rDn' csr.. ?ressed ratios D52' 3 Llr~(N~=2s5~ !h.ll he`r. cs~?ute rstios 0025 Lr~`.- (255)-25j !Else wsuld over,low, 'ro.t 255 tt, -255.
DD2 5 c311 d3 ta ( nDrne ~ ~ 4,, 3i, ' ) DD27 ca'l data~basis,'l,,') OD28 do 4 n-l,lD24 0529 4 buff ~n)~D
003D eall orn(fcb,bufs,1,D O) Itr,3ke blue lnvisible; ~reen ~r11 be C teo, s;ue to ~S~, rnss' o' :he titne, exeept in rEEZE.
OD31 1~ 434 N=l, 255, 2 OD32 5_~CI(N I)-D
DD;3 C34 &~CI3(1~' D)~' IC~3ER 51C~5 FUL!~:IG~ ~ crr. PRDDU2T.
OD34 C~U LLrr(Fc8~LL~N~2~4~o~o) iGreen oe.'s neg. input frr~n ~h3n. 2 OD35 C~:~ Ll~(r~,LUl'i~,2,2,D,O) OD35 C~'L.L L~.~(F~3,6,2,0,D) !Green ~ channel I - ch~nn31 2 OC37 C~LL LUi(F~3,LU~T,4,2,û,0) !Rea~y so use rs'ios 0038 ChLL l.L.'-(F~3,Ll~L~:,4,1,0,0) ! A~,, olr' 6e~uenti~` prscluc~s OD35 IDO 20 N=1,512 !I~. rnaps -255 ts D, O to 127, and 255 tD 255 no4 D 1~ (N-l ) ~2 DOql IF~I~.35tN-255~.L_.:~SEt-~S) ~127 OD;2 2D buf f (N) ~P
OD43 ~L IFM(FC ,bQFf,512,511,0,0,0) !C!~:, 2Z20/aO, ~ro LO~D O~ E Cl~rLY
C US_ GR~EN CCt~`5~512, ~n't ~ffe~t norlrla, fee~ck of gre~n,EYPIFts.
DD44 C311 const~fcb,O,S12,0,D,O) DOq5 do 30 j=1,256 OD46 3D buff(j)~

;1 ;=257,Sll 004R 31 b~Elj)~255 !~,ite referexe forces ~11 shite-e~cee~ers b~ck to 255.
0049 c ~1 if~tEcD,buE,0,511,Q.O,O) Q050 ~?1? 19, ' ' ,7 C
C L~~ s~E ~sE aF C~F20LLI~ ~r~ FILE ~ND OPEN IT
QOS~ r~l i dr~o( ' EYEFI.Y' ,InFut) 0052 IT~O
0053 i f ~ In?ut .r,e. 1) qo to 2300 0054 type ,'En,er da~3 file Eor EY-dFLY.' OCSS ~Cce?t l9,(c2:nfillj),j-1,29) OQS6 ~ 2200 Nal 2~3 OD57 IF !C~IL~N; .?~E . ' ' ) Il YP~l OC58 2200 C~rINllE
OQ~ 2300 IF~I~P.~Q.O~ call d~ta~ccm'll,'~l,S~yeny.dat') OQSQ OP9~' (UN~2, N~ fil,T~PEa'OLD',R~DONLY;
0061 re~ 12,9) nhl,~N~,cf~lse,Eul~D3,r~1nch,r~axch C --InFut I hierarchies, rrnd~controlled swltch, r~ult1-color/1 se?3r~ n C--s~itch, un~sr~d variable C~ALSE, r~in~roL~, r~ximuc c~lor ch~nnel l's.
D062 n~leta~
Do53 if ~Euljn~.eq.l) næmlet~7 Q06q re.~d t2,19) In~rne~ ramlct,29) Q065 iE jRAE30~.NE.l~ GO ro 35Q
ca~
C ~ RAh~ ~, for e~ch 'hierarchy of level~, C here inr,ur the ju3l? sl2e, ~ jU~?s, threshold, 1 ~aths, ~ th.,t jurp ahea~, C and c~e'ficients for poolin~ fracti~n.
c OOSS do 300 n=l,r~i OQ67 read ~2,3g) jum~n) ,nho~n) ,i~?~n) ,n?a~n) ,ah~ln) OQS8 3DD reDd 12,69~ chlln),ùhiln),ehi!n) QC69 GO ~Q 400 C ' ` ~ 51=:~ =
C CF~, F~R C~TROL~ MPS, re2d in the slr31~ threshold, poolirq C fr~ctlon coeff.~ients, and the x,y, com?onents of ev~.~
C ju~?-vector, and conventlon~lly deflne eYactly one 'hierDrchy"
c of one '?ath~ m?~sed of th~se v~ctors.
C

QD70 35Q read ~2, 9) i nteps 0071 re~ ~2,69) C,D,E
C072 do 360 r~l,nhi OQ73 36Q read ~2,9) jcx~n),jcy~n) 0374 ~1ulFnhi OD75 ~1 il~I for controlled series-l lor-3 psth of variable jum?-sizes!
007$ NPA~
OG77 :00 CL05E ~ ~Z
C~
C ~rERtsINE niE ~IES OF INF~ Wi E~ IM~S
C

C~
Q078 if ~input.ne.l) 9D tD 2400 0079 ty,oe 509,7,(n~e~ nlet,29) OD80 1(fulir.. e~.1) ty; e , 'Will store sec iors of color Sr-Q~e:
I <ret~ D~ enter inp~t IY~name. ' oQ3~ a.ce?~ gc, (tna~e(i~ ,i~mlet,29) D082 OD q9D j=n~et,29 0083 if(trK~me( j1.ne.' ') go tD 495 CD5. 49D continue ooQ~ 1 0 240~
Do56 495 do q9~ j-ramlet,29 oD87 ~ 9~ n~me(j)-tr~ne(7) !~e allD~. t~rrlin~' inpu: oi image-n~me OQ89 2400 a~*~nd~-O
ODB9 rx~nd~O
OO9D If (fuliTtg.n~.l) go so 2500 0091 rs~and~inch C092 rb ~nd~xch oQ93 cell d~tJ(namou~.,'1,sl,3~,') OD9C c~ll deta(n.~Dus(9) ,name(7)) 0095 dD 2410 ~3,30 0096 i'(r~Du~ .eq.' ',or.r~ut(li.~$.'.') 9~ tD ~411 0097 2410 contin~e 0095 stop 'Disk f ilen20n~ tr~ lorg . ' 0099 2'11 if ~r~rn~u~ q.'.') go so 2415 QlQD næ~ut(l)~'P' ~?rocessed im~ges get5 'P~ the eæ Df its n~T,e.
olol gD tD 2500 0102 2~!5 do 2~20 r~29,1+1,-1 OlQ3 2420 n~oout~n)=namou~(n-l) Ib~np letters fDrward tD m~k~ ro~ fDr ~P"
0104 n~u',(l)~'P' 0105 2500 oDntiT~uc OlD6 ~IYrO~secr~slD.) C I~OP Qr DIFFE~7~ COLaR SEP~
0107 DO IODD IE~A~,~^Q~.~ 1 ! ! ! dD mult;pl ~ CDl~rs ir, on~ rUT:! ! ! ! !
0105 c~ll on~s(buff,l,-l) !h~ (~11 bits Dn) in c~lan. O.
C

C IDO? OF Sl~CESSlV~ ~IER~ ~ rE15 FaF RA~Y PAh~5 C OF DIFFEi~ ~SIZE
OID9 d~ SOO IFI~l,N~' OllD if jfuling.e~3.1) nQme(5)~eib~'0' !input channel picked frcr, imT~ge 0111 c~ll d2zz(r~me) 0112 ty?e ~, !skip ~lor~, SD ~V~X etc. does-.'t ovcr-write mcss~geS.
0113 if ~rznd~n.ne.l) g~ to 150 C ~
C INl~I:~LIZ BIEQ~ SPE~IF~C C~S~Q (jump 5iz~, C possiole vector c~r.ponznts, poolirg raction coz'ficienta~
e~
0114 ju~psz=jun(ihi) Q115 IFA~512ilU?~P52 !far e~e of s=reen (recruirino insulation) sU~rts bere.
0116 nu~nhD ( i hi ) 0117 inteF~iep(ihi) OllB np~thsYnpa~lhi) 0119 ~e~cah~(ihi~
0120 c--chi ( ihi) 0121 ~i(ihi) ~:~L8~1~3~

0ol232 ~~ ','J~:r.p -.i~e (r~x 8j, ~ o8 ho~, inte?e: thresh, ~PAT~E;, ` A~--~D (4i' ,f7.2) ' 0124 T~_ 35,J`~.~s,l~M,inte?~ i?A~5,~AD
0125 if(n~3-hs 1e o~ npat.'~sl 0~6 ir(checd le.0) I!lheal?50.3 0i27 eyr~ ~,'Real coef'i~ient~ C,D,E to calc. frac~i~n weightlng olc SEO?R' 0123 t~? 79,n;~,n~sn 0ol230 ~ ,'8it bu~.trn ~ to p~use and consider e~rly exlt.' 0131 TYPE 19 IS?D e or~iurd 0132 5 IF(~0t'~$Z.'T.0 .t~. JU~SZ.GT.5P4) JUMPSZe8 013. D0 6 N50~ 7 C'3~ ID:i:Y.(!i)oI~'D;?~ )bJUt1;$2 8 3i 6 Illlr.Y(~ ~lRY~ SZ
C ID0? 0' I~I'LT:;LE P~TH5 ~. Gl LI; ~ n~ L
C ~irs~, ini~icli~e ?c~specific pramaters) C ___ _ _e-~ ~=s5esesr~
0136 150 do 200 n~ ,npcths 0137 laseccl~ran(irl,ir2)"6.
0138 ~;X;0 0139 ~rO
0140 i(n?.eq.1.Dr.ihi.gt.l) qo tD 8 0141 crl] ones(b~,1,-1) Iw~ (a11 Dits Dn) in chan. 0.
Cee--~55--~
C LClO? OF I~iDIVlDLP~L J~ ehe iNNER W;~I~I~S! !
C _ __ __ =
Q142 ~ ID0 N~0?51~N,I~
ir(~'~P.eq.l) GO 10 16 C=- ~
C C~tior,ally, vie~` p-ogress Dn this ?ath so ~r.
Ceses -_ 0143 if ~E~R.e~.l) call rbutn(cb, jcbut,ix,iy) 014 ~ CA' ' PaU;~` ( F; E~, B~, IY., ~Y) 0145 !~E8~D lusuolly will nD- need to cle~r butt~7n next t~ne.
Dl 46 IF ( IBU~ .NE . ~ 0 16 0147 I~ ,'BI~ but~,Dn U tD res~De pr3cess, C to see origincl, D tD ex~t.' 014B ~YPF 19 !E?ace for\4rd 0149 7 call ~aiti~2) 0150 ca~ rb~tn(fcb,iblr,ix,iy) 0151 if (ibu-.. ne.l.~.ibu .ne.2.nn~.}BU;.~:.3.on~.1bu~.. ne.4.an~.ibut.n~.8) go tD 7 0152 if (ibLr .eq.8) call exit 0153 if(ibLr~.eq.2.cr.Bl~r.E0.3) oo to 15 015: call Itont(fc~,ibu',4,0,D) !shou chi~nnel C vs. channel 2 01SS o~ tD 7 0155 15 NEElr~l ~bu. here, rnust cleor, ~n3 n~ is tDD e~rly.
0157 c 11 Itcnt(c~,1,4,0,0) 01$8 16 if(ibut.ne.3) ~o to 17 0159 tyFe ','Enter E;~ElErJ~sk: I shif~, 2 rDtiD, 4 newprr~uct' 0160 occe?~. 9,mfrn2e 0161 17 IF(RA)~.1a:. 1) G0 ~0 11 C __ _ C IF i~., X:I~SI~& SPE-IFIC Y~ F~ I~IIS J~IP:

3~3 C (pro~2billty ~ A~n of veerirq of eourse of lDSt jl~lp) 0~ 62 r3n~y- ran ( irl, lr2) D~ E3 ~aVE~LPSlOE) 0164 IE(r~y.le.~eaè~ GO ~ 10 O~ 65 D166 I~(randy~L~ (I.~ahe2~n.~ IDÇ:I,--I
0167 I'D~ )(LASi~IOEL,E) 0163 10 U~ 'E
0159 ~-IDIRXi~VE) 0170 ~V~'=IDI~Y(~V~) 0171 ~o to 12 C ~==~_ C ~, ~ V~ frcm lnput list, an~ flnd C lt5 ~2gn.tu~e 3ne th2t of complemen~ry wr~p-arour~ e~ge.
C

C~ ~
D I 72 11 r~--j ex ( nhop) 0173 r~jey(nhop) 0174 junps~:~ats jr~x) ,abs~vy)) Cl75 lfar~e512-jul~psz Dl 76 12 T~X~I~X~
0~ 7 ~ NY
0178 ~PE 4S,l~rX,q~rY,nhop !lype a^cun~ulate~ e~xcurslor, of ju~ps so fDr.
C
C E~TE;ND SE~--I~L F'RO~JS!!!!!II~!~I!!!!!!I~I
C Next five lines ~e31nnirq ~r,th ~C~l.. - pl!rfon7 tne vi~l functions:
C 1) hlft Oriqln~ aqe C 2) F~æ it eo Intermediate C~aruiel C 3) Feæ ~k Patio o' st.iftæ ~rigir,al to the Int~rmediate Channel C 4) Ene!:le the Contribu.ors tD the lntermæirte Roduct C S) F~ed th~ lntermedi3te ProdL~t b3ck to et,e IntermF~iate Ch~nnPI
C

017g C~!!. L~q ~F-B,4,1,0,0) !Ena~le blue fro~ ct~sn. 2, latest im~e-shiFt.
0180 C~LL fee~2 (b~',l,1,-l,l,M~:,MDVY) 0181 if(iar~rr~roze,').ne 0) e311 freeze(2 2~ !(op~ional oause for a look~
Cle2 CALL FE;~--r;(FC~,2,2,-l,C,0,0,0) IStoré 'ratios- in ch3nnel 1.
0183 C~L L~--(FCB,3,4,0,0) JFc~ fro~r (lol S~?r~+~loq r2tio) 0184 if(iand(mfr~ze,2).ne.0) e ~1 freeze(2,4) IRED uses ratio LU~
- I (op~.iona' p2~e for a look) oleS C~LL ~ee~2(buff,4,1,-1,0,0,0) 0186 lf(i~(mfroze,4).ne.D) call freeze(2,2) I(op~,ion21 p2use for a look) Cle7 call ltcnt(fcb,1,4,0 0~ ~ne to ~tct, ehe pro~ress of secuential pr~.
C IP ~D>D ina~en~e just srre3c in positive directir~n,so C \de mLst s3fekeep Ist rieveral rows fre~ ~rap-around by rp~rlng C the CiD prod~t (ehan. 0) ehere into ehe pcrt of new product C (fe~ edge) ehat WWTD stheruise ctiange the~, when r~lC, ne~ ere C ~veraged.
C IE' ~OVEKO, trice-ver~. ~t is, s~fekeep high2r ro~s of old C ineo lo~r of Dew.
C _ . . . _ C --~2vision ~IN, 2~16/80, 11 pn. New r. e~d is E~ster and will han~le C cases ~7ere both Pl~; ~rd M~Y are nonzero. Wrap-aroLnd z~nes are C r~arke~ with ones in blt-plane zero (detail lost anyway ~th C~.50!) C Tnese r~arked are~s effect 100~ ~iqh~ ln~ of old pro~uct, thanks C to 434-lr,op inltlali2ir,g of ~AC~ ~rrays, a~re, 3~3 C .~
c C Cl~hr7 u~r ~0CIS F~ and then !S~RX h~P-Ar~v"~ NES
C

C---------~ __ ___5--~ __-- _ __.~5--~5i=:! _5___ ~__.
01_8 rsll fd~k(fCb,2,3,I,1,0,0,1) I~ero rJu~. plinne 0 oL ch5-s 0,1 DlBS Ir~JU~:X~ 0~ 56) G3 T0 1035 C ~ ar a-G~r~'' ~7rries for rniq~ae Jlrpsz~rr-r~en ~jdth)~2!
01 gC dG I D3D i=, J
C!9i r~rd-r~ IJ (directiGn) Dn> ro~, for ~,.ch Y--chanqe 3s );ey, 019 2 i f ( ~ . er; . I j rnoved-rrrqv Dl9; if(rw.ed.eq.0) 9G tc lD3C Ino 17rk neede~
01 q~ i LrGrn=D
Clr' if(rove'.lt.D! iLrGrr~ifsr olr~ itr~ sr-i~rr. n 019- cal; rr.arkl(h1~r,j~m,r~ss,ifrr~T.,ito,l,j,2) 017i 103C c7ntim~?
01~9 1035 rontinL)e 02D0 rnv~rr~vy D201 r.~ rs (-r~x~8~8 D202 i (r~x.lt.~vx: ~ri~8 D2D3 ip~srx-(-r~x) I~leec` t~ scroll' the ne~ se;L~e.7ti~1 p.rdL)ct C beore it can be ~averaqed ~jth o'd! a7d fe' back tG ~han~ D.
C 59, os in feeom2, e br~ dD"n every X scrG`; ir,to a ricrGll and C a ~neq~tive) righ~arc D-7 pixe! oi'set: d~7 the c'ise- first.
C~
C D-tG-7 prxe; 'ofset- fDr precise unscr711ing e-- . e5~ __ _ _ D2D~ i(;~x.e7.0) go to 33 02CS c~l; imace(fc:~,b~f,S12-i~,D,517~ipx,1,2,-l,;,l,D,D,D,D,l) D2CS C~ l~rr(F~,2,2,D,D) D2;77 CBll fdxk~fcb,2,2,-l,l,ipx D,D) 020r' cal; ltcnt~:b,5,2,D,D) !Gét sreen reod~ for next f~TlO ~:~ir,.
D2n9 c~ saGi(ic~-~bl:~L~o~o~s:2oipx~l~2~ D~rbo~DlD) C ~esidLal) UN-SC~'~L~; OF ISI~UL'r~ ~ PF~
C ~
C2`0 33 call s:r71(frb,m~,my,2,0 D) Cr~ ~
C

cC S~ U? ca~:3~I10t~ 7I.L.r, Y~r'~7.0'JS P7O~5 Cr~
^v211 if(n70p.s-.l.ar)d.d.e~.D.~nd.e.n5.D.) 9D to 435 IS F~'~S arr~ys re5dy.
D2;2 ohsoldp~n~op,nL~m) D213 if ~oh.lt.D) ohsD.
D214 iL~sh.gt.l.) oh-l.
0215 ro (D) ~oh U216 co Il )cl .-oh 0217 do 34 ici0,1 021 a ds 34 n=0, '~54, 2 0217 34 s:rach(n,ic)cco(ic)^n D22D CD 35 ic=0,1 0,2; 35 c21~ fcb,scrach~D,I~) ,I,25ic 0,U) D222 435 c~ll ltcnt~cb~3~l~o~o) I81ue wili be ~ w2jghted ov7. of old,new S~Qpp5 C
C _ __ ____~_ ~_~.__.. ~
C ~er~ back Csroination to S~'TI~l Pr70W~ C,Jii:.~L
D223 36 CAll fdr.,ck(cb,1,1,-1,1,0,0,0) 3~)g3~3 C ~
C F~-ZERO ~e scr~lli.~ re3is:er nee3f3ed for ~LT!i-~3LL3'?~-22C c211 r~crol(fcb,D,0,2,0,0) IT~a33LE, i~ don't res~3t.
22~ 3~3~3 C~T~3~
C~3 ~
226 If ~n,~3-hs.e5.1) go to 200 Ino ~3v~r29irlg t,~ dol C c~Lnr~ F~ESULS F~ OF l'~Y P~n!S, ~nd C ~optlonally) cvD~ate It to singlo r,~ost recent pr3th 227 TF(!~.~.O~ call dl3ez('2",ey~.te 226 do It~ rr~O,25$
.29 r3~l_h(n,1)~(tn;~ noat(n))~'10i~(np) !aver;,~3e ol cld ,~3ths 23Q 133Q m~p;3th(n,Q)~noat(n)~Eloat(np) !no~,3 p~th C W~ig~t Dccu~ulated A~. UP by ~3 f,~ctor of (h'~l) coDlpqre~`3 to n-:~ .sir,31r pDt3l.
231 dc 135 rr-D, I
232 185 call 1~3t(fcb,avp3~h(0,n) ,4,2~n,0,C3) Imr3ke hEL3 ~ updDtt!d aver&~e.
233 c211 Itcnt(fcb,3,4,0 0~
234 call dtck(Ecb,4,2,-i,1 D3,D,D) 235 C~L! LL~r(F~3~LLlrL3~ 4~3 0,0) !or cY2Din2tion, llne2r; lDter restore h~T
236 c211 ltcnt(fcb 2,4,0,Q) 237 t~e 59,7,np,nr 2333 ipushd~D
239 i~ait~O
240 19Q ra~l ~itlc(2) 241 i~it=i~ir31t~1 242 if(ipus'nd.ff3ØDnd.iwr31t.gt.30) go t~ 19; !so prog cDn rL~n severrl C I p2ths on Its o~rn.
243 c~'l rbutn(fcb,ibut,lx,ly~
244 if(ibut.ne.l.2rd.ibu-.. ne.2.Dnd.ibu,.ne.4.Dnd.ib~3t.n~.8~ 9~ ~o 190 245 if(ibu-.ec3.e) 9~ to 19S
24E c~,ll ltcnt(fcb,lb~3t,4,0,0) ?47 iF~2sh-i=l !Cnce y~3 mr3ke D reoues~, must push D to go on.
24e go to 190 '' ?q9 1975 c 11 rbutn(fcb,icb~At,lx,ly) !Ciear button 25Q crll I da ~s ( ' 2, 1,, eye, te~rn; 1 ' 25~ if(np.e~.npaells) go to 2DD
?52 c~ll ltcnt(fcb,l,4,0,0 ?53 cDll lu-.(fcb,lu~rDt,4,2,0,0~ Irltios fror~ ch~n I to hed enDblet?, Pi~
?54 200 crJntinue ~'~7~ r*~t~*--*
?55 TF (h~;' .?.'.1~ t;O 7'0 SOD
?56 c&ll ltcnt(fcb,2,4,0,0) !57 call fd~ck(fcb,4,1,-l,l,D,D,o) !Fer~ red (avg. at this level) C to channel D for nex- Icvel !SB call Itcr.t(fcb,1,4,D,O) ~59 call lut(fc~,lu rat,4,2,D,D) Ira~ios Er3n ch~n IG> Red enDbled, ~;A~'!
~60 50Q continue C~ t ~ r ~
'61 if(fulim~.ne.l) go to 1000 iskip s~orlng oE results.
62 nr~ut (3~
53 if(ibL~.gt.D) na~out j3)o'_' I~E: a red record i5 written to ?~1 C f ile, other records au~ment or update &n OID ~ile.
~64 nrxDout(7)~ib3nd~'D' call daz2(namout) ~66 IDOO continue C ~ - - t ~ - ~ t - t ~ t ~

I-ll 3~

267 t~ ,'SIme ela~ ',secr~s~timeO),'secor~s.' 2 63 9 FCF~T ~16 i S ~
2~ r~80~1) 270 39 ~ ',S ~4~5,~7.2) 271 49 FC~T ('+~X,1~1~`,21S,' hop 9',iS) 272 S9 formz~ ,dl,'P~th',13,' 1~: button A~last, ~a~. r~'',13, 1' C-orJgin~l, D~;O ~1',/) 273 69 frrT3a~ unctlon 19 C ~ nh~p ~E'~nlp~' ,13,'), nlp vary1rq I kx 1 ^~o' ,13,' o~ eacb pdtn.') 276 509 fofmat (Ix,al,' <retu~lV tD process:',26~ or enter lDISI~, 1 inpJt-ilr., ~ ) 277 E~D

~7 ~ ~

t ~ 3 ~ 5 ~ , 0 ~ t l r~ -_ _ _ ,~ ._ _ _ ~ ~ ~ r ~ ~ 0 ;~ t tA r ~

~ ~ ~--0r ~nr t t ~tnt L, cl ~ ~ ~ i@~,~ r ~0 C~ - X ~ ~ X Ln ~ T--2 n ~ y ~ ~

W rv ~ N rv N-- N N---- N rv N ~ ~ 1~ tv G 3 D '3 ~ ._ D t, v tD tD 3 t3 O ~ ~ 2 ~ 7 tr v~ C ~. t~ _ L~; ~ r~ vr ~ t~ L~ trnL r ~ ~ C

N _ N N N r~) r..) ~ N N N rv tv t~ t~ tD t3 t3 O t~ t~ tD t3 tL ~3 --' ~
r~ c~ r r ~ 3 ~ vq ~ t~ C~ Cq ~ 0 N

- _ N N v rv rv rv r~ rv N rv ~

r -- C~ ~ rv ~ 6 ~I r~ ~ 2 r ~ 2 ~2 Iv _ N r~ rv N rv rv rv rv rv r7 ~ r~q i~

r~ '~ ~ 0 ~q ~ LV ~ ~ rV r ~n ~D 0 0 3~

., ,~, _, ~ _ ~. ~ _ r w ~ ~ ~ 7 ~ C r C ~ ~ ~! c 7 _ c ~ ~ ~ r~
O~ r _ _ _ r r .~ D r ~ r N
_ _ _ _ w _-- r' ," r~ * ,~ r~ r~ ,~, r~, r~ r~, r~i r~
r3 rG r,~ ,. G ,~ r~ , r~l 3 ~ ~
e t3 ~ L3 LJr ~'1 0 r ul ~ r r I r ~3 ~ ~3 r r - - _ 3 ' r~ r~ r, r r c a a - r ~ IX rv ~ L ,~ r`~l ~ æ ~ ~v ~ Q ~ ~S " ~ r' ~ LJ ~ ~I L~, ~ L~ L~ r~ r~ Lr/ ~ L3 w rJ r~ rJ rJ ~ ~ r, ~ ~ rJ ~ r~ r~ rJ r.
L N - L'` - L3 L~ L~ ~ LD ~ ~ r Q L C ~ r r r r r C G ~ ~ ~ X L ~

_ _ _ W 1,~ _ W .7 _ rr r~ C ri _ _ G Ln r. " ~, C 7 ~ r ~ n r~ L~ r, 7 Ln 3~ Y Ln r, C 8 r, 7 r r m W n rJ ,~W, Lr~ ~ rW~ r L~ rJ
` ''' 'L- Lr ~ 1~ Ln ~, . n ~, r c, t~ 9 3 c -~ r ~ r L~ r. r r, ~7 rJ r ~ L;
_ _ _ ____ ~~ L3 U
r N _ ~ w - 7 -n r .L r ~ ,G~ ~L? 8 '? '' '~ 5 r~

_ _ _ _ L~ _ _ _ 7J
Q ~ ~ ~
rJ N r~ C~ L9 .~ O r~

rJ Q~ ~ r~ ? _ cn W W-- D
3 ~-- -- -- G g ~ L~
W ~ O ~ a ~ ~j .1 ~3 .f ~- C~

r ~ 2'3 c ? ~,~

h~ ~ IjJ ~ ~ rJ

~r ~ ~r~ r~,3,, ~ 8 æ C7 r~ 3, ~ 5 ' Y r ~ ~rn ~ q~,~

V~3~3 .~ ~

..1 7 r~ Vl n '~, llB~'~3~
Sv~RO''-~' C:~SS (F3, cua, cc~, c~, VRSRX, RE~) sB?~R;rvTD1E RE~-S t)~ h'RI1'5 17~ Ca~SA~S RE:;ISSER.
n~-E~,R FCB(11, CCt~, CC~X;, CU;R, Vi~l~C, REW
C~ SH' ADDlT~v COt~S~.~X Ir~EI~ BY rnE Evl.Ur SvP~ PRCI''ESSOR
C~r;C - r.`: ~'`DITri'E CCt~S,~;' BSE~v BY rnE GPv-E~' 5Vu Pi:OCV~SOR
CC~n - rlE ~D ITI~v CONSSh.'~q U~E~ BY rB RE3' SU~ PRBCESSOR
F~D - O S'`IPLIES h'RSl`E, I IY~PLSES FWD~
RvB~ J~E 1'L~'-R ~FCB, C;)' OR, CXAN!v~ elSP, BIPlFl~1, PIX~vtFP~ XI~RN, ZEiO) INTEGER F;3~1), Cl)vOR~ CH~ BITP, B~PIFM, PIXB~FF
IN~2v-B EXi'ER~, 2ERO, VRTRT
S'v?vR!~B~l~YE S~' RE~ D hr~ lE FEEYIACK LOi''P C0~ROL h~v~i ~::a - A?t Il~"2C.ER ~`' JSED F`OR S'Y5'-E', 112P~ E~, l~C~`SA7`IC~';
COLD~ -- COrvOR SrL2CS WV~D
4 ~'~ RT'D, 2 e> GÆ3~ > 5Ill.E
C}~3WL - A BET M~P S_L2CSlh~:; ~L- C~VL FCP 7~1E D--S'TI~Slv~
I ~> IPAGE vn 2 le> IY~ I
4 '~ ~M~tiE 2 .

~63B4 '> I1~5AtE 14 -3276E '> IM~h.E IS ~C;RA~Ir5) EvITP -- h BIS M~? SI ''C--D~; SRE BvIS PI~ZES Xl RrAD~ITE
tOP~V,2~ L~ -1, I.E. PvLL Sll'S. ~E XCEPIIOI: TO rr,15 RBLE IS hliEI~' h~lI`-I?I; I~ r.rE G~PXICS CHA!~EI.
B~.?IF~ - G IMPLI~`S USE IFM, I IMP;IES BYPP5S IFM~
PIXv~F - PIX2L Or~S~, ~ C~ N C~ 7.
EX~ ' - I IM~.IES EXI~iAL INPL~, I.E. Ll'v;I7'IZER
22i'~0 - 0 ~> Nv~S~L~ FED BAC~ ALL 0'5 tUSFvr ~:fi BLANNNG rB2 Elr51'IN~TSv~' Cl;AN.) 5U3PVvr~ , (FSa, I~P, STARS, COJNI, PACR, VPI'RX, RE~D) SL13RVv~-NE ~ )~3 OR h'RlTE A SE~ON 0~ SaE l~v'S FvNCT1Ci~J Y~CRY (IFM) .
~;R PCB~l), MAPtl), START, CW~tS, PP'~-lt, V~R5C, READ
I)~2GR BIl~S tl6), P51~5 M~ - A ~CUJt~ h~P3 ARRAY TO P~ESWC~AD; r~E IF?/, C~S
S~ - rr~E F~SSlIC~` tZERG P~EL) I~ S.~ IPM ~riEvRE I~EE
SRANS.rER I5 SO SS~
COU~'S - S~E NI~BE~l O- I~l 1~'rS TC TRP~SFER
PEAD - O I~?LIES h.~TE, I IMPLIE5 P
PP._I~ ~ I I~ IES PP~D Mv~ ~SFER

`` I-16 U~38 SUaRO'~n~E Ih~i;E (F~, PtYELS~
XJNI'r, '-~11', t``PTXEL, DIRECr, 2 C~t~iL, PLA?~S~
3 Ph~D, eY?lF?~" B`.'r, ~ .~, e JRS
4 VRTR~ ~ F~) SUail;)'~D~ RE~tS Cfi h~rRES D~t~,r D~TA.
~'llX;ER F'C3t~) ,' PIXELS(I) ~ Yl~'l'r~ YtNI'r, I~;PIXEL, DIRE~
t~.~`E.'Efi CM~, O~`L, P'L.~S
1~--C~.. R PAC~ B~'PIF7`'., eYTE~ AD-h~ VlLORD, E~'URS~ T~ ~ fiEAD
PIYE~ - A~ --r~Ei% ~aRAY r;~ RE--EIVE/CC~ AD~ E T~b. E D~TA
Y...;ll - ,~ E tl:` ~E FIRS- PI~E, r~N~FE~D (;1 F~L) `.IR t'- - ~ Y Ct)~:;SiI~;P.T_ LF Tl~ FIRS~ F;YEL 'rR~SFF;~i2 (O P~EL) N';B'Y_;. - Tl~ ~0~~ NUi'3E~ O} PiY. LS 1'0 ~SF~
D;P~EC~ - D 't~.PLIES P.-J,E~hfiIl'E PR~cEETtINv 7~ 5'LE PIG~, I l~;?_;FS F~-D,~\-.~rE P~XEE IN6 D3~'~t ~'_ ~ A B'~ M~P SELECIIh'P~ rriE C~`;EL~S) TO P~EPDJWRITE:
3 ~> 1~ D
2 ~ ~ I ~AC~ I
4 -> IM~r 2 E'rc.
I63R~ -> 1.:5A~_ 14 -3276E ~> I.V~GE lS (GRA~IICS) ~-.i~.: h'RIlI~; ON'LY~ 'rHESE CO::CE r.by 3r C~3INED
~:) 'r~I~E ~r SAV.: DATA D~O 7'h':) O;. r.~E c~rLS.
F~R EXAMF E, ~r~D~L = -3275E hOULD ~SE~ Cii~`~LS
l, 3, 6 15.
P~ES - A BIT l~.P SE~FCr~; TdE Brr P~'ES ~1) P~D/~RIT, t~vP~LLY -l l IE. ALL EITS. 1~_ EXCEP;'IC~' TO THIS
RUL IS h`'B3; h'ifi;~'ING I~ r~E ~?.SICS C}U~L.
PAC~ - O IMr_1E_ 3 BYTE~r~3l 1 ll.?LlE~ 2 SYTES/W3?D
B~.~IFV - G l~-ILES USE IFM, I I~SP' IES BY?ASS IF~', BYTE - D IMr'LlES ?~t~R~ALI I IMPLIES 3 PIXELS/BYTE~
I . E. ~ BIl~bXY D~TA.
tTE - XI~'IT M535T 3E A MU_TIP E :~ 8 - 0 I?~PLIES 1~3R~L, I I~1-7LIES Tt;AT TdE, D~TA D~
r.E~3RY(S) IS OR'ED T~ TdE DAI'A FRESE3~r'D FRt~S
THE CtY~rER A~:3 T~ RESLFT IS S7~ rdE ME~tSRY(S).
~ t~"'TE ~ USED hrd~' WRIT3' X; 3t~LY! !
BURS~ - G IM?LIES ~t~AL ~SFEr., l I!PLIES BU?ST (E~ST) XFE~ii.
~ N~E BI~E~l MUST BE SPECIFTED.
V~;7~X - 0 IM?LIES h'fiT'FE A~'YTIM-, 3 I~PLIES h~ITE ~ING VER.^ICAL P~5~LCE tN'Y.
R_Jlt - G IMPLIES h'.RITE, I ItSPLI_S READ.

~ 9~ .~

SlJ::~R(:)i.~.~'' ~ ~F~D~ D, COLC3R, Vr~R~C~ R~
S'J3RWI:[~ 10 RE~ iR;I`_ AN E:NIIRE Ol~P~ ~C~ON l'.E71~1Y !OFM) IN,^~--~D F'C9 (~ ~D(I 02~) ~ Cl ' OR, t'RlR~C, RE~D
~,~D - A 1024 h~pr A~AY ~C, P3~1VL/CON~ArJ~' n~ C~. CO~EN' S
COLOR - A CO;)E INrlCA~ h7UCB O~'Y ~ RE~ E:
-> aLu~
2 -~ GREEN
4 -~ RED
h'r~N h~.'l`lNG ON'Y, 'nlE5E CO_~_S ~U.Y aE
C~J-131NED TO h~2TE rBE SA~.E ~ IN''O ~tt`O
0~ ~..R-E OFI~'S. r~i; EX~?LE, COiJ~S WW'D
h~ `r_` B.~J rB~ BLIIE AND Fi~D O~.'S
P~ - O II~D` IES l~,~.I?E, I II`LD~.IES RE~D.
SU3;tO'~NE SCRO' (F5, S~R~:, E~_h'OLY, C~l~`', t~:, P~D!
SU3RO`JiINE P~ S OR 1-~7UTF~S A S'CROIL Ca~.qROL F~ S~Efi.
~'R ~D(I), S-ROL):, SCRO'Y, CH~, t3~R~C, R~D
SCR:)IY - ~`~' IN'rE..~:R SP~ NG ~BE SCRO! t 1~; OE~FSFI~
r~ x (B~IZ~:T~' ) D;REC'?ION
(N~r ~. O!:LY BO~ ;Z~ S"ROLLS n~ iNCR:~ OE
E PI~LS AD_ F355r3L- Tr~E 3 LEAS. SlG. B~S
3r' SCR3LX ~!: IGN3REI)) SC20;Y - A~' INSES_. SPEOIF~'ING rrE SCROLL31~ru 3EFSE~
3X` Ir Y r~.3CA' ) DI~lION
C~"t!~' - A Et. ~P SELECSING rAE CHAI ~--L(S) ~G R~4D~W2'~E:
~ -> 3M,;C O
2 -~ IMAGE 1 4 -> IMAGE 2 16354 -> IMA~iE 14 -32'7D8 ~~ AGr 15 (GR~.ICS) WAEN h'R3,~ ONLY, rAESE C3~5 M~Y Er C3!;3INED
Tv hR:rrE rBE SAM_ D;-~A 3)~SO ~h'O OR 1~73~LE CBA~ 'ELS.
P32 EXPMPLE, C~L r -3275E WOULD 1~ CBANNELS
3~ ~ 15 ~D - D 31~'~PLIEE WRIl'E, 1 3MPLIES 3~D.
SU3R3'.J:'3NE R3U;N (F~, E~N, X, Y) Rour ~r~ T3 RE~ !3V~J~` W~RD A~
CUR;32 P35ITlOh`

3~ 3~

SU~RO.~T~ LTi ~~-- (P5, ~S~;K, COLOR, ~rl~ ~ P.EP,D) SUaRO;1:`2N5 TO R~ R h'RTTE Tl~: Lur ~IASK (S) 2?~ 2 ), M~ , CI~ PE:~
I~SK - A~` ~E`C?_R h~OSE EiTl' ~P DErE~1~5 h7iiCIl I~K UP TA!~ 5 ME E~ 9L~ 'SABLEI) LSE~ ~ 1 ~> E~ }3LE 0~ L`qORY
. . . ErC .
COLOR - ~ CO~ q)TCA.ll~G h'HIC)~ LUi` 1~5K TO R~D/h?~ITE:
~ - ~3L~IE

4 - ~ED
7 - R~GRE~3LU::
- O Il-'PLT~:S h--r~ 2 I~'IP.IES READ
SU~rR.'~D~ Lur (F1~ P, CO'~i, ~ .~, VRlRTC, READ) S'J~;RO'~IN TO R~3~YRiSE ~; Et~.lRE I~OK-UP .-ABLE (LUI) R FCB~l), I~P(256), CC)LO;~ L, ~I~TC, R~
Y.~P - A 255 h'aPD A~Y TO RECrT~/C~ ' ~ LU;' C{~h~S
CO:~I - A COD_ Ih'3ICA il~, h'lilCI~ LUr TO P~li/l~
1 -~ Bl~
2 -~ GR~Eh' 4 -~ PFD
W'.~ h~ITINB ONLY, 1~5E ca~s Y~.Y 3E
COMSi~;D TO h'2iTE TdE SAb'E IP.l'~ I~O Tr~O
t~ T~E JJ~'S. F~ ~.YPLE, COLOPrS h'W' n h~RIl'E a~n Ti. BLUE A~'D P~ LUT~S.
C~ ', - A E,T r3~P SELC'rII~ IIIE ~EL~5) TO T~J~lTE:
I --> IM~ O
2 -~ Ir~E I
4 -> IMi;E 2 E~C
36384 -> IMAGE ~4 -3276E -> Ir~G~ 15 ~GR~'.ICS) W.-~ ' h~TT~ ONLY, IR`SE C~S r~.Y E~E CCr~ST7.~ED
TO h'P.IlE ~1' S~YE ~I'A 1~'--0 7W0 01~ Y.~ CH~EL5.
F~ E~5PLE, C~L ~ -3275E WOULD ME~N CH~':'L5 1, 3, L 15.
- O IY2LIES h'Rm, I IM~PLIES il~).

3~,, f~R~ ~-~US '22 _i~ 12 ;2 19 21~1:iff3 P$~ I
~1 F~ ~TP ~R
E~tl ~ubrùu~nr r~rr~ltbuff numl ln l Inrl n~wl Ichr~n Idlr IIEST) C t'lLrti<l ~5 C sp9clO~-pUrpr,C~, ur~ncrnl stJbrout~n~ ~0 C ror~ o 512~by~rl_1t1 srctl~n of ~tTOpO ~n two chonnrl~
c t~flE~i C buif- ~cro~ch buff~ o~ 1224 ~n~tp~rs nordetT bu subrrt~no C l ~nal- 15 thc~ l~t rDw or colurm C nLwl- Is tts nr~w ls- rov rr colur~n C ~ct~n- 1~ ~h~ rolrt~ont ~'~R_Et chonnrtl r~ott~
C ~d r~ r~ rDr rDt~C I fnr cclumn~
C Id~Tt- d~st~no~rn chrnnr~l r t~ ml~nq D~ e drt~oul~ t~ -IC~:N
C ItEG~ '9qTE ~tSE IP J~SI-ICI~ r1 rnd ~hll~lhT--l ~hon onl u C orc ~r: rf onr~s ~ dro~rt In o ~o~o o o~t o~ ~r_ 4~t/E3;~
81~2 Intctg4r fcb(E) Y~ l) YD~a I ) bUt~
~33 ~ntrtgrr ouff(loe4) ~a~ co, I ,. r9ChL '7"oc~ ) ~5 or~ 1 n-l la24 C~D I ~ t(n)--l E~37 Dutrt~l D~a~3 ~ù~r~ Idlr l!~a9 Jor:t~ldOS~
1- Jd~r 1- 2 or Jdlr gt 1 ) Jdlr~
Dall ~-(tolslnur~l~n e) nr B) but~-a !~n th~r oppllcot~Dn th~ cond~t~Dr c ~ ~uTf ~clLtnt to o~D~t ItTlTX nDn--rtult~pl o of c3!Etal2 ~f ~ lD~St rq E~) Jd~tt--Ichsn a2!3 npois-lnuml 1~1-31)~2 ~Etl4 lrst-rùdlnt~tl ~n ;12) Ja15 ~fllr~t rq e) irrt-32 lS dr S n~d 1 B217 ItDl n i--a æla s Y~n)~
a~l9 IGO-l ElBZe IE!~
E~21 11~-1 ee22 If tllnrl gt n~wl) ~DO tD æ
23 IgD npor~
el224 ~ Dnt~l eP2--D ~nc----1 !t~urt f l Tp Drdrr tD Ut~Dld oorl t, D~.ur~rltc~ Df par:l prrhrps C !n~rdad In I ctrr chunhtT
ie25 Zt dc 3Et t ~D land Inc Eaa7 num 32 a~2e ~fl~ cq npo~tT) nutn~lr~t ~3 x 11 l-Jdlr)-lInrTlfft-l~
aa3e ~co~ l-Jd~r )-nr~wl~ 1 )t~32 c~el31 co~l ~r~tda~f~c buf~ Y~ta) sl~1) 51~U1 Jù~r ~hon l l l bu~r a a a ~!) V~t~

e}~ ~l`I,Idor.~ nc Icbon.rr r.;~l n~ l~nRI) coll Ir~o I ~-ct,bu~ (tO,~ot~,s~2~r~Jr~ Ja~t~ l bJ~a,a,~
t~33 30 ront Ir~ur~
~334 rR~urn OE~S ~nù

a3~3 O X r ~ r r ~ rl ~ ~ N _ j~ N N N N ~ N N 1~1 N ~ r r ~ ~ N 0 ~ N
N ~ r c~--N N ~ ~ N N N N ;~ ~ N
r = r~ ? Y ~ , v, ~ r ~ r p r p p u~

N N N N ~ ~
,V~ V~, ~ ~ Z~

D N N N j~ ~
~05~

~3 N N N ~ ~

I-2~

~T}~ ~D ~BRC~I~S ~E..r~:D

TOT~ S~E 1410:P~E~ 27^t 3~

rr~ T ~E; U~2-S1 12 42 ~q 2~}~
r~e FTh ~7R f~L~D~
e~lal QII~J711E i~lM21bur~ICAO;~ 57~JTrJ~P~JYPlrM~IX~lv) C ~12 ~o~d~ col~r IC1~ Ih4,G~,Bi~ D ~hc~nLI IrEST, o~isLt~ln~ Sho C Irapc ll~'u,' 1~ rrlgln) b~ SX ond IY I~DI~IED C~79 t~ wgrh C fo~ \y by u~na IPX~ In FDE1X fcr x-oi-~ut~ not a~ol~lblr.
C b!J b. I
C --tD po~ch tho tlngl~-Dl~ol In 11 oc~ursc~l elol~d rd ~nDnths o~r C I---I'~ In faadbrc~ I~O~D5 ~DI-rol blook c~lurn~ In ~hr.
C Ir~r~t l~on ~f tho d~elr~tlon uhlch con b~
C dD~ o~ trt I ng ~ I
C ~112 USO~ E~FF ~uppl ~od to klurL In th~sr fDW colvlntS, ~ I th 2 c5 ~
C Th~ IIS ~croll crr~-ntl~n~ opply to thlo coll lhot ~91, c IX~e rorol I s ar Irogr to thc ~7, ond IY~fo scrDI I r W~`v 7hD
C OpDrC~l~r UDVld praducc In~DrDstlng bvt norrallu undo~lroblc r-~vlt~
C ~ ST rhrnn~l eDntrlbut~d ~o lcr~-r~ b~Cov~ tho ecrol Ird IcoLr~l C IrlDrG r~v bL dls-Dr~d dvrlnr ~udbr~c~ ~n lnrriT To o~ld ~uoh probl~ms C ~h~nnol E7~P crn b~ fud rllsu't~ ~o u~d r,n ~o lE57 ~n~
C COS~ D- ~ chDnn~ l cDntr ~but In~q t~ d ul f r--q~J~r-~ un ITE~P
C chonn~l bo o.olloblo I' I7EIP ~ nDt n~d~d ~c~ It t~ -1 c rOr n~ Ci~LS r~Y ~n con-~lbuto tc C lcoLr-lR Gunurcl l~ng ~uld c~c~lr~oln currDnt ~orrl 1~ fr~ ol I
C rDn~rlbutlng chonnDI~ ond Indl~lduGII~I upd~L ond r~n Doch r~no C b'tPlF11 1~ e to u-~ Dr 1 to bupo-~ ~h~ Input un~lon ~rlDry C rbts Ir~q) thot ony col I t~ fucd boc~ o COIDr tD ~loh th~
C dlr~tl~or oontrlùutD- n~u-t In~ludo on I~D chonn-l DlnC~ th~
C d~D~ r ~honnol ~on ~lot b~ ~orol lod ~ID ~h~ orlolnal du~gn C o- ~cr~lllng ~ho d-~tlnctlDn Inv~r~21u foll~ b~our~o It aOO5 ~el7 C of f ~t i~/~dbock !
5 d~ nust bL ~6-J-mDdl 11l 9)1 ~SJ2 butos lonc, ond ~rrt on o ~Drd bDundor~
t~2 l~rr'~ Et~P~r~t rt t~ r.~,rrrr.D~O
~3 IntL2rr buff~255~ llru~ ort Dn b~t~ b~ndory 00i~4 htnl-~7E~P
e~dS tFth'JIl L~ a~ Xr~
e~0 C~LL LT9~tT~FC-c,C~lRlB lclLoRle d S~pG 1~ cn~rIb- ,entrib ~57 JY ~ ~ Y
e~ Ip--O-i~rd 0E1;?q rr~nDdl 1~ 61 101e l f ( m ~r~ . Et I po t,o 2a r~al l JY~ 3 ~t Cd312 If lJ~ It 1~) Jx~ Et IFDB31C Dff~rt ~ t bo rloht~rd ~-fcCtlv~lu<2) e213 ~?~f ~Y-t~
0al~ 2e ( f ( Itunrc I t el CfLL SCR~t FC--D, JX, lY, C~T,RIB, fd El) r~rdl S c~L ~ ~ F0B, I CI~LCR 2~tl, -1 I En IFM L, 0, e ) rd~

l~a15 If tlpYorf Dq 21 p~ so ZS !n~ "corrqct~rn~ naod~d so rlll In Ir,~t odg-0al7 coll Inooo(~cb,buit,51Z-lp~ùff,f~ 517rlp~ùt-,1,2~ nvml,-1,1,1,t3,B,0,0,1) E!l)IB CAl Fll~<lFCE,lCllC~,2~1UUIl,-1 E;~rPtFa~\p~r,ff,t~,BI
B1315 c~ll Iroa~tf~b buff~Qlrdlsl2~lpxo~f~ll2~ nun~ ll0lBlE~ lB) t~û 25 ~ t~rlp gq a)'t~o o 3~3 t~l t~ 5C~tFCE,el,B,C~(lRit3,0,01 ~-2 9~ ~ù lBC3 t~23 ~ I~v 5 J-1,25~;
t~24 5 ~v~ i ~ J )-J- ~
eezs C~l LTo~T~rca,~2#rt2-lDE5T)~B~ t) !~osd curr-ns rf~ or 1~5T CDI~l`
B~25 ~ LLntFCB,bv-f,21l~l2-iDESTI,2l1~iTaP,8,e) ILr,~d lln~rJr LUT ~o~ FIIECK
BBZ7 cr~LL LTt~TtFCb 2il~l~rP 2~1lt2-'tl~5Tl,B,S) t~3 coll ~orol~tc~'J~ ,2C~t~rp,l~,a) 1~29 ~11 F~BO~IFCB ~12-I~5T),21~'t~,-1,1,51,2,8) t~3 coll ~ùrrl(~b,2,t~,2~X~tr~p,9,~) 0a31 CrrLL LTC~lFt`Ji ~5K,23n5t2-It~) I,B) IRr~r,rr, cr,lor ~ prv~lou~ ~r~
ae3~ ET~
t#33 ~

~1@~4t,., ~j N N ~} N N ~ ~ ~ 3 r ~ 8 ~ S~

3N~ 5 ~ 8 ~ x ~ ~ ~ D

i~ tUNN ~ ~ N

p r ~ p r ~n ~3 ~ N ~i N

N

8 ~
~ N ~ ~
~ ~b~ ~

" I-26 tOTfL ~: f~L~RT~D ~ ~21aa ~i41 PP I~I~CTtO~S (~tl D
~1: F~Z, ~3: 6 ~ZZ ~ kes ~ rharaceer strin~ cc~3nding ~ tr~nsfer of 512x512 ir~ ~sapara~ions betw~en dis~ ,~nd ~efrestI m~y.
~ITIC -ve:~i sp~ified nu~ber of 6Cths of a r,econd FrE-ZE -p3use~s to ellov ex~rln~tion of content,s of s~ified ci bnn~l.
D~ -e,~ proc~ssing.
-fill ~n entire ilr~a3e plan~ or panes w!etl on~.
-lnitir~lize en r~rrrly vith a gi~en sl~numeti: literal.
~IC -p3~ r~ ~ait r~veral ræcc,r~s report b~ch If there ls InplJt durino, th~t eir~e AL -r~rr.;ali2~s all funceion~l t~1bles In the ~st~.

3i~3 II-l APPENDTX II
A preferred color masking stage 58 of FIGUP~E 3 performs linear transformations of all ~hree sets of processed-image information Erom the image processors 20, 22 and 24~ A first operation calculates the average of the combi.ned outputs from the three processors 20, 22 and 24. Wi~h most images t.his average has an approximately unifor~ dis~ribution of values from (0) to (255)-A second operation, performed in parallel withthe first, calculates the difference between the outputs of the information processors 20 and 22, which process the long wa-~elength image information and the middle wavelength informatiGn, respectively. The resultant color difference between the two outputs is represented on a scale of (0) to (255). At a given pixel, all colors with equal output from processors 20 and 22 will have a value of (128) on this scaleO The difference value at most pixels in the image falls in the vicinity of this middle value, i.e. (128). Only the most saturated colors approach the extreme values of (0) ~o (255). In fact, if the color detecting system 14 has overlap in its spectral sensitivity functions it will cause the total range of difference values to be limited to values higher than (~) and lower than ~255). Based on the range of difference values found in a variety of typical images, one can expand the limited range found in the particular hardware system to fill the range of potential color differences. This can be implemented by taking the range of values determined in taking the long wavelength minus middle w~velength difference, ln this case at difference values from (80) to (175), and linearly expand ng that distribution to values that range from (0) to (255).

A third operation of this color masking stage 58, performed in parallel with the firs~ two, is the combination of half of the long-wavelength information from processor 20 plus half of the middle-wavelength information from processor 22, minus the shor~-wavelength information rom processor 24. qhe resultant represents the color differences between two other color bands in the color space. As in the second operation, the characteristic range of values associated with the particular system can be linearly expanded to fill the range of potential color differences~

After this expansion of the color difference properties, the color masking stage recombines the expanded transformations of the image by computing inverse transforms of the original transforms. This produces a cclor-enhanced image that is sent to the exposure control stage 60.

The initial transformation and the color expansion described above are fully detailed by the following listing of a program termed ZOW~E and the data ~ile it requires, termed ZQWIE.DAT. The listing and the data file are in the same language and for the same equipment as described for Appendix I.

[ 0.33 0.33 0.33] R L
C 0.50 -0.50 0.00] x G - Cl -127.5 ~-0.25 -0.25 0.~0~ B C2 -127.5 q ~

After placing L, Cl, and C2 into the three memories, ZOWIE then expands Cl and C2 ~y the amount specified in ZOWI~.~AT. The inverse transform proceeds in a similar manner.

Beginning with L, as is, and the t~o e~randed color difference channels, each minus 127.5, the mathematical inverse of the above matrix is multiplicatively applied:

[ 1.00 1.00 ~0.67 [ 1.00 -1.00 ~0.67~
~ 1.00 0.00 1.33]

In performing this second transformation, the only new requirement concerns results more extreme than the storage range (0) to (~25), whlch can occur because of the expansion. Such values are replaced with the appropriate limit, i.e~ either (0) or (255)o (~he remainder of this page is intentionally blank.) ~Jg~ ~

3~

Il-4 PLU~ VD2-S1 10:41:D0 ~4-AU~80 2 0~ - r . E-rN /?R: ~WR
OD0,1 PRO~R~ ZC~IE
DDD2 INSEOER F~(6), HHFFER10:1023) OD03 II~E~,ER BUFI~.,~0:255), BU~G(0:255!, BUTBMY(D:255) O0Dti INiEG_R LR,I~IR,LOU;,8IO~
DDDS OPE~UhlT I, ~re~(111,4~2:hIt.~? ,~?E OLD ,R~NLY) DDD6 IODD FO;v~1~215) OD07 RE~D~l,lDD0),Ii~,HIR,2~Ur,KlOU.
ODD~ LO 13 1--0, LR
0DG9 13 BUFL PS~I)=Lt)UT
DD10 DD 14 1=HIR,255 0C11 14 BU~ I)=HIOUS
0012 )2NC=FL~)A, (HIOJI-LOUE)/FLOAl (HIR-IR) D013 DO IS I LR~8rR
0D14 15 BU~LJtl(l)-Nl~(~lP.l ~LOJr)~FLOAI ~I-LR) Xl~C) ODIS ~::OO ,1000) ,LR,~IR,LC);i?,8101JT
0016 DO 213 IC0,LR
0 17 213 BUFRM8~I)=LOUT
D01e DO 2Iti le.lR.255 DDl9 214 BUF2MSiI)=HlOUT
OD2D X3NC~FL~1 (HlOUr-LO~)/F~A~(B3R-LR) 0D21 DO 215 I=lR.HIR
OD22 215 BU~tl)~s1.~(FLOA?(LOUS)+FLCA?(3-LR)~XlNC) oD23 RE~D~l,lDDD) ,Lh,HlR,LOUS,8IOU~`
0024 DO 3 3 I~C, iR
DD25 313 BJ~SY I)=LOUI
OD26 1~ 314 7=HIR,255 CD27 314 HU~.~Y~I)=HIOU?
OD2e XI1; =FiQAr~HlO`~-LOU~)/FLOhl~HIR-LR) OD2~ DB 315 I=LR,HIR
0D3D 315 BUF~ ~i)=NIN~lFLa~ LOU~)+FLt~?~l-LR)~XINC) C
C

C ~F~I IM~GE ?C Ll~:.llJANCE-CHRQ .Il;~CE CHRt~ ANCE SPACE
C
C

0D31 CArL Na~L(FC~3 !3U2P~R,3,0) OD32 CAI~ L?C~(~,;,7,D,D) C

C I~IIS PROSRA~, E~ ~ ?A7 E AN I~E ~ RE~RLSH Wl?H PEB (R~, C GREEI~` (G), ~D BLU~ (E) SEPA~.I'I~S I~ CH~ELS D,1,2, ~SPEC~ELY~
C A?3~, REPU~E I~ BY ~ IEh' iMA~iE IN W~IICH CHA~L D h'ILL CC-~A:~
C 'L'~'INbN~- (L), E:ED~EC AS
CC L= (1/3~^(R~) C C}~L I WILL C~q~A~; SHt COLOR ~IFFEFU3~E RE3 ~IIN~JS a~EN
C 5 ~ O ?HE RA~;E D ~O 25S:
C CIC (1/2~ ^ (255~-G~
C

C Cj~EL 2 WILL C~ mE COLOR DIFFEREI~E BLUE MINI35 YEL~
C SCAIr TO S.'KE R~E O SO 255:
C C~ ( 1/4 ) ~ ( 5 1 ~2E~- t~+~, ) ) C
C

C S~E PRX~ WILL NOS RE~IIRE ANY ~tG ~ S?~INJ OF I~$~5, ~T
C L~3E ~ LDOX UP SAE~ S, CC ISIP~'T RE~ISSER, IFrS, AND FEE~ I;.
C

C
C

~ IJ~ ~

3~

C Sl~:P 1. RPLACE CH~EL 0 BY L
C033 DO 10 I~0,255 t)034 10 3UFFERII)'~I
OD3S CALL L~'r(Fl~,BLrFFl R,4,7,0,0) 0036 DO 2D i~0,765 C~37 20 3llRFE~ ~(.rLC~2~'rtI)i3.) no3s C~LL IFM(r~3,BUFF;~,û,766,0,0,0) 0039 CALL FDB:X(Frr;,4,1,255,0,0,0,0) C
cC S . EP 2 . REPI~CE C~tEL I 3Y Cl C ~.,/E: Sl~-E CH~W:'L C }~S B~ RPLArD BY L, hlE USE A DiFr~3~ FOR`;ll~A
C P~ CCYS~ ;; Cl:
cC Cl~l/2)~t255+3L-2G--B) c ~r.: ~E LOO~ UP TA3LE FCF~ nrL IMAGE 1~,' l'H ~HE M~:IMU,`' SE;LEL' ~E
C ~ MrLY r3~2~L) h'lLL BE !;CALED ~3 l7SE PS MU-H Or Il'S Rpl~r p5 C P25513L_. lr.tE OTrL=R ~P3LE5 h~LL BE SC,~LED APPROP3IP.~ELY. ~r C MULl'IP_ICATIV-- 5'_PI E FA~ MUST, ~ COU~SE, 3r ~E SAME FCR Al~
C TRREE L5~r~ UP ~A~LES. THr trRlO'~ At~DlTiVE OrFS_'r; (l:SED T3 M~E
C 7~E LC~ES. OU'P~ B_ ~?D TO ~rlE 3E~Nh'Il~G or THE LOOH U? I'ABt' C O'~P~, -256) I~ILL 3_ CO-'PE}~SA~D FOR Wl~ TF C~IS--~A~ REC,IS'rER
C THIS RE~,IS~R hriLL ALSO BE IJSED TO Y~3LE l'HE 255 OVERL' ' Ln )t-IvL
C Ct25.--~. ~E hr.iD_ ~nII~; h'iLL ~' Br FED SAt:X TO T~E C TO 2;5 C R2.N0E OF REFr~SH rHROU~L: 'rHE I~ ` FLIN^~IO~; ME?:~Y ~IF7`.).
C IJSLY F0)R C2.
C --~
cC I~' LINS WIrB rrL A301iE, h'E P~W;IITE rrtr ~ULA FC~ C1:
C C1'~3~C)-~2L~--4/3)G~(-2/i)E~17û) OOqO DO 30 1~ 0,2,5 00U 30 BUrF~ER(I) ~ 2~1 ~ 256 0042 CALL LU'r(F~3,BUFFE~,2,1,0,03 ûD43 DO 4(i 1=D,255 ûD44 4û 3UFFER(I)=NINI'(-4.~'FLlY~rtI)/3.) + 84 OD45 C~LL LUl`(FC3,BUFFER,2,2,0,û) OD46 1~ 50 1:0,255 00~ 7 5D 31~FER ( I ) =N11~ 2 . ~ I ) /3 . ) ~ B6 OD4B CALL L~n'iFCB,BUFFEfi,2,4,û,û) 004~ C~LL Ca~'ST(FCB,0,C26,0~û,D) 0ûSD DD 6û I~c0~ 340 D0 51 60 B UFFER ( I ) -N8~ (3 . '~ FI~T ( I ) /4 . ) 0052 CALL IF~.(~,3UFFER,D, 341,0,D,03 C053 CALL Fl~cY(FCa,2~2,255,û,0,'D~O) C ---- -- - .~
cC S~EP 3. k:PLACE C~W~VEL 2 BY C2 C AS F~?R C1~ ~tE, WE MLlSl` R9~-~TF PC~ P~R C2 USING Tl~ aUA)i~ITIES
C ~AT 5TILL EXrS~ I~ R ~ESY.: L, C1, 3. ~C1 IS N~3T RE~UI?~D HERE.) C C2=~3/4)''~17D~L) C
C

i 6 OCS~ DD ?Q 3=0,255 ooS 5 7D BuFFER ~
D DS Ç CAU LL~ ( FCB, BVFFER, 1, 4, D, O ) OOS7 X 8D I ~D, 255 0058 8C BUFFER;I)--I
D0SS CALL LVr(F3,B~ ,I,1,0,D) 0O60 CALL L~?~r(F_B,S,1,D,O~
DD61 C~U C~5r(FC8,17D,D,O,D,O~
OD62 DO 9D 3~D 34D
DD63 SD BL~FER(3~ rt3.~FLt~ 4 ) DD6q CALL 1.~.(FCB,B~FEB,D,3ql,0,0,D) DD65 C~LL FDscl~(Fcs~l~q~2s5~D~o~D~D) DD6r C~LL ~ ~(F~,6V.~.`E~,3,O~
c ~
c C NCh' C~NGE SL~7PE OE LU~.3~-E CH~EL
OD67 C~LL LV~r(FC8,8L~LL~.,4.1,0,D) DD68 CALL F~:~(FCB,4,1,2S5,1,D,D,D~
DD 69 C~LL LUI ( ~FCE, 3UFF~G, 2, 2, 0, D~
DD70 CAU ~K(.rCs,2,2,2sj,1,D,D,D~
OD71 C~U LVr(F~,BUF~,1,4,0,0 0072 C~l FlBrl((F B,1,4,2SS,1,0,D-0 D073 S~? ~2~IE~
DD74 ~D
7~ E.17Al file tl~at speci~ios magnitude 7f c310r m~s3cing trensf~:nati~
o 255 LU~CE CHAN~EL rN ~L~', B3GH~ (2IS~
D 255 OL'r ~LCW, HIG~ t2IS~
8D 175 R~E~ C7;h~EL n~ (LC~, HIG~I~ (2IS~
D 25S cur ~L~ ~G~;) (215~
8D 175 BLUE-YELL~' C~EL D~ (L~ In~H) (21S) 0 25S our (LO~ (2IS~

J

FI~rR~A.`~ PLUS V72-S1 17:41:00 14-AI~-80 t~ lE. F 11~` ~qR: A~l,fh~
PRXR~V SE~ICY~S
~;_.~3ER N~r SIZE ATI~;aUl ES
3 SC03-1 003D30 780 RS~`,I,CCI~LCL
2SPI:Ar~ DDOllD 3$ ~,O,~,LCL
3SlD~T~ ODO;4 2 177 Rv ,D,CC~,LCL
4 5~7~5 DC7032 18DS Rh,~:) CC~,LCL
SE-EY~PS OOODD4 2 RW,D C~Y,LCL
VMIA9 L~rS
~SE ~?r AD ~ESS l~-YE q~?E ~ SS ?J;~SE I YPE ~ SS l~.` E T~?- A~R.:
P.IO T I-2 4-D07D22 i~IR 1-2 q-OD7016 1 I~t2 4-OD7D24 I.OU; 172 4-OD7 2 4-OG7D14~I~C Rs4 4-OD702S
~AY;
TYPE AD ~ESS SIZE UI~SIO~S
EllP~ 2 4-OD6D!'i DDlODO 256 (D:255 B ~EER 1~2 4--ODDQ14 DD400D 1024 (O:lD23 E~ LL.~. J~24-OD4D14 ODIODD 256 (D:255) B~!~; 1 2 4-Dcsclti OOIODD 256 ~D:255) F`1 i3 1~2 q-ODoDDD DDD014 6 ~6) l~qE'5 LPt3E! )~ESS 1,L3EL All)RESS L~i3EL ~ESS L~3-L ~ ESE
a~ 3.3 ~ 14 ~ 15 2D ~ 3D e,. 4D s~ SD
6D ~ 7D ~ SD * 9D
213 ~ 214 ~ 215 ~ 313 314 Yr 315a~ IDDD 3-DODDOD
FU~C~S ~;D SU3RC~JIII~ES REFEi~ ED
Cal~15~ FEB~ l L~ L~ ~L OP-J~S SNII~
~AL 5PACE All~:R~D - 0127qD 2E~0D
Z~I_~05P~Z~IE/LI:6 APPEI~DIX I I I
A preferred exposure control stage 60 of FIGURE 3 uses the output from the color masking stage 58 and computes an op~imal use of the characteristic dynamic ranges of each color component of the color display, in this case Polaroid SX~70 Time Zero brand instant film. Exposure control functions that map -the numerical outputs of the FIGURE 3 system into those w~lich produ~e ,he desired representation on this film are calculated. ~hese transformations are small because of the powerful lmprovements already incorporated in ~he image processing stages 20, 22 and 24, and in the color masXina stage ~ This final exposure control stage is thus designed to fit the limited dynamic range of outputs from processors 20, 22 and 24 more closely to the pax~icular dynamic range of the display device, e.g.
the photographic fi~m being used. The numerical transformation performed in one control stage 60 for the above-noted film is shown in the accornpanying plot of FIGURE III-l~ The transformation is identical for the red, green, and blue channels.

3~3 255 _ /

O UT PUT OF
EXPOSlJRE C128 _ /
t:;ONTROL 1--STAG' 6C) ~64 f O 64 ~2~ Ig~ 255 Dl~ IT 1 N

INPUT TO E~:XF'OSURE CONTROL STA(iE: 60 ~OUl PUT C)F COLOR MASK S~AGE 58) fi~ur~ m ~

Claims (86)

THE EMBODIMENTS OF THE INVENTION IN WHICH AN EXCLUSIVE
PROPERTY OR PRIVILEGE IS CLAIMED ARE DEFINED AS FOLLOWS:

1. Apparatus for producing an image of a subject which com-prises A. means for detecting radiance ratios between different areas of said subject and producing a first lightness-determining quantity in response to each such ratio, B. means for effecting said ratio detection for each area of said subject a multiple number of times with other areas of said subject which are at different locations on said subject relative to that area, C. means for combining each first lightness-determining quantity with a second lightness-determining quantity associated with one area in that ratio and replacing the second lightness-determining quantity associated with another area in that ratio in response thereto, and D. means for producing an image to the subject in which the lightness of the respective image areas is determined by the last replacement values of said second lightness-determining quan-tities.

2. Image processing apparatus for determining a field of accumulating measures of image lightness in response to informa-tion identifying optical radiance associated with arrayed sections of an image field, said apparatus having the improvement comprising A. means for sequentially determining a comparative mea-sure of the radiance information for each segmental area of said image field relative to said information for each of plural other segmental areas, said means (i) providing a new intermediate value for each such measure in response to the product of a ratio function of the radiance information associated with each first-named segmental area and with each second-named segmental area and of a like mea-sure previously determined for the second-named segmental area, and (ii) determining a sequentially new value of each said measure in response to a selectively weighted averaging of said new intermediate value and a like measure previously deter-mined for said first-named segmental area, and B. means for providing a replacement value of the pre-viously determined measure for each first-named segmental area in response to said sequentially-determined new value, thereby to determine each measure in the field in response to an accumula-ting succession of said measures.

3. Image processing apparatus according to claim 2 further characterized in that said measure-determining means includes means for determining different ones of each said plurality of measures for first-named and second-named areas that correspond to spatially different sections of said image field, where said spatial difference includes at least a difference in size of or in separation between first-named and second-named areas for which that measure is being determined.

4. Image processing apparatus for determining a field of accumulating measures of image lightness in response to informa-tion identifying optical radiance associated with an image field, said apparatus having the improvement comprising A. means for sequentially determining a comparative mea-sure of the radiance information for each segmental area of said image field relative to said information for each of plural other segmental areas of that field, said means (i) providing a new intermediate value of each such comparative measure in response to the product of a ratio function of the radiance information associated with each first-named seg-mental area and with each second-named segmental area and of a like measure previously determined for the second-named segmental area, and further providing each such intermediate value with reference to a selected condition of said product, and (ii) determining a sequentially new value of each said comparative measure in response to a selectively weighted averag-ing of said new intermediate value and a like measure previously determined for said first-named segmental area, and B. means for replacing the previously determined measure for each first-named segmental area in response to said sequen-tially-determined new value, thereby to determine each measure in the field in response to an accumulating succession of said comparative measures.

5. Image processing apparatus according to claim 4 further characterized in that said measure determining means includes means for determining said measure for each first-named area rela-tive to each of a set of second-named other areas, where said set includes areas at selected different image-field locations relative to said first-named area.

6. Image processing apparatus according to claim 4 further characterized in that said measure determining means includes means for effecting a determination of said measure for each of a selected plurality of areas prior to effecting a further such determination for an area of that plurality.

7. Image processing apparatus for determining information corresponding to image lightness in response to radiance-identify-ing information for a selected image field, said apparatus having the improvement comprising A. means for representing the radiance-identifying infor-mation for each of selected segmental areas of the viewing field, B. means for determining a selected comparison measure between said identifying information for each segmental area and said information for another segmental area, and for determining therefrom and from a previously-determined lightness-identifying quantity for each latter segmental area a newly-determined light-ness-identifying quantity for each former segmental area, C. means for effecting a selected multiple of said deter-minations sequentially and between segmental areas that correspond to differently-spaced locations in said field of view, and D. means for producing said lightness-information for said image field in response to said multiple determinations.

8. Image processing apparatus according to claim 7 having the further improvement in which said means for effecting said determinations sequentially includes means for applying each newly-determined quantity of one determination as a previously-deter-mined quantity in a subsequent determination.

9. Image processing apparatus according to claim 7 having the further improvement in which said means for effecting said determinations with differently-spaced locations includes means for ordering said determinations according to the magnitude of the spacing between locations.

10. Image processing apparatus according to claim 9 having the further improvement in which said means for ordering said determinations effects said ordering with determinations between areas of largest spacing being performed first.

11. Image processing apparatus according to claim 7 having the further improvement in which said means for determining includes means for operating on a set of said identifying information that is responsive to said selected image field to effect one said determination for each segmental area of the image field prior to effecting a further determination for any such segmental area.

12, Image processing apparatus according to claim 7 having the further improvement in which said means for effecting multiple determinations is further arranged to produce for each segmental area a lightness-identifying quantity that is responsive to radiance-identifying information for substantially every other segmental area of the image field.

13. Image processing apparatus for determining information corresponding to image lightness in response to radiance-identifying information for a selected image field, said apparatus having the improvement comprising A. means for representing said radiance-identifying information for each of selected segmental areas of the viewing field, B. means for determining a selected comparison measure between said identifying information for each segmental area and said information for another segmental area, and for determining therefrom and from a previously determined lightness-identifying quantity for each latter segmental area a newly-determined lightness-identifying quantity for each former segmental area, said means for determining including means for effecting a determination of said measure for each of a selected plurality of areas prior to effecting a further such determination for an area of that plurality, C. means for effecting a selected multiple of said determinations for said plurality of areas sequentially, different ones of said sequential determinations being between segmental areas having a spatial parameter different from that of the areas of other such determinations, and for applying each newly-determined quantity of one determination as a previously-determined quantity in a subsequent determination, and further including means for ordering said determinations according to the magnitude of said spatial parameter, and D. means for producing said lightness-information for said image field in response to said multiple determinations.

14. Lightness-imaging apparatus having means for providing information identifying optical radiance associated with arrayed sections of a selected image field, said apparatus further comprising A. means for selectively grouping segmental areas of said image field a selected number of times, different ones of at least some of said groupings involving areas having at least one spatial parameter different from other groupings of areas, B. means for providing, for each grouping of segmental areas, at least one measure of visually significant transition in said radiance information between areas of that grouping, said measures being with reference to a selected lightness condition, and C. means for determining image lightness for each arrayed section of the image field in response to a plurality of said measures, at least some of which are provided for groupings which differ from one another in at least one spatial parameter.

15. Imaging apparatus according to claim 14 in which each segmental area has a geometrical center, and in which said means for grouping provides different ones of at least some of said groupings among areas having a different spacing parameter between the geometrical centers thereof.

16. Imaging apparatus according to claim 14 in which each segmental area has a geometrical center, and in which said means for grouping provides different ones of at least some of said groupings among areas having a different spacing dimension between the geometrical centers thereof.

17. Imaging apparatus according to claim 14 in which each segmental area has a geometrical center, and in which said means for grouping provides different ones of at least some of said groupings among areas having a different spacing direction between the geometrical centers thereof.

18. Imaging apparatus according to claim 14 in which said means for grouping provides different ones of at least some of said groupings among areas having different separations between the edges thereof.

19. Imaging apparatus according to claim 14 in which said means for grouping provides different ones of at least some of said groupings among segmental areas having a size different from the sizes of segmental areas in other groupings.

20. Imaging apparatus according to claim 14 in which said means for grouping provides at least a first of said groupings with areas of uniform size and provides different ones of such groupings among areas having different separations in terms of at least one spatial parameter selected from the parameters of distance and of direction.

21. Imaging apparatus according to claim 14 A. further comprising means for assigning each segmental area an initializing value of said measure, B. in which said measure-providing means includes means for providing each said measure in response to the product of a ratio function of the radiance information associated with grouped areas and the measure already assigned to a first of said grouped areas, and C. further comprising means for replacing the measure assigned to a second of said grouped areas in response to said product-responsive measure.

22. Imaging apparatus according to claim 14 further including means for resetting said product-responsive measure with refer-ence to said selected lightness condition.

23. Imaging apparatus according to claim 21 in which said measure-providing means includes means for imposing a threshold on said ratio function.

24. Imaging apparatus according to claim 21 further compri-sing means for retaining the measure assigned to said second of said grouped areas, in lieu of said replacing, for each said second grouped area which is grouped with an area located beyond said image field.

25. Imaging apparatus according to claim 14 in which said measure-providing means includes means for resetting at least selected ones of said measures with reference to said selected lightness condition.

26. Imaging apparatus according to claim 14 in which said determining means determines image lightness in response to an arithmetic averaging function of plural ones of said measures provided for groupings which differ from one another in at least one spatial parameter.

27. Imaging apparatus according to claim 14 in which said measure-providing means includes means for providing at least some said measures in sequence with one another and for providing sequentially-successive measures in response, at least partially, to a preceding measure for an area of a grouping, thereby to deter-mine said image lightness in response to an accumulating succes-sion of said measures.

28. Imaging apparatus according to claim 27 A. in which said means for grouping provides groupings which differ by a magnitude parameter, and B. in which said measure-providing means includes means for providing said sequential measures ordered between groupings of largest spatial parameter and groupings of smallest parameter.

29. Imaging apparatus according to claim 14 in which said measure-providing means includes A. means for sequentially providing different ones of said measures of visually-significant transition in radiance in-formation for the same segmental area, B. means for assigning each segmental area an initiali-zing prior value of said measure, C. means for providing an intermediate value of each said measure in response to the product of a ratio function of the radiance information associated with grouped areas and the measure already assigned to a first of said grouped areas, D. means for providing each said measure in response to a selectively weighted averaging of an intermediate value pro-vided sequentially previously for said first of said grouped areas and of the prior measure for a second of said grouped areas, and E. means for replacing the prior measure for said second of said grouped areas in response to said averaging, thereby to determine image lightness in response to an accumulating succes-sion of said measures.

30. In lightness-imaging apparatus having (i) means for providing information identifying optical radiance associated with each arrayed section of a selec-ted image field, (ii) means for selectively pairing segmental areas of said image field a selected number of times, each said pairing being of segmental areas of identical configuration and size, (iii) means for providing, for each pairing of seg-mental areas, at least one comparative measure of said radiance information at the paired areas, and (iv) means for resetting each said measure with reference to a selected limit condition, the improvement comprising means for determining image lightness for each arrayed section of the image field in response to a plurality of said reset measures, at least some of which are provided for pairings which differ from one another in at least one spatial parameter.

31. In imaging apparatus according to claim 30, the further improvement A. comprising means for assigning each segmental area an initializing value of said measure, B. in which said measure-providing means includes means for providing each said measure in response to the product of a ratio function of the radiance information associated with paired areas and the measure already assigned to a first of said paired areas, and C. further comprising means for assigning a replacement value to the measure assigned to a second of said paired areas in response to said product-responsive measure.

32. In imaging apparatus according to claim 30, the further improvement in which said means for pairing segmental areas in-cludes means for providing different ones of said pairs of said areas with different spacing parameters ordered in said sequence from pairings of the largest parameter to pairings of the smallest parameter.

33. In imaging apparatus according to claim 30, the further improvement including A. memory means arranged for storing said radiance-iden-tifying information, B. scroll means arranged with said memory means for pro-viding said selective pairings of segmental areas, C. signal transformation means for providing said reset-ting of measures and for providing selectively transformed informa-tion in said memory means, and D. adder means arranged with said memory means, said scroll means and said transformation means for providing said comparative measures in response to selectively scrolled and transformed information in said memory means, and for providing an arithmetic averaging of plural reset measures, at least some of which are provided for pairings which differ from one another in at least one spatial parameter.

34. In imaging apparatus according to claim 30, the further improvement A. in which said means for pairing segmental areas inclu-des means for providing said number of pairs sequentially, B. comprising means for assigning each segmental area an initializing value of said measure, and C. in which said measure-providing means includes (1) means for providing an intermediate value of each said measure in response to the product of a ratio function of the radiance information associated with paired areas and the measure already assigned to a first area of each pair, (2) means for providing each said measure in res-ponse to a selectively weighted averaging of the last previously provided intermediate value for said first area of each pair and the last previously provided prior measure for a second area of each pair, and (3) means for providing a replacement for the prior measure for each second area in response to said averaging, there-by to determine image lightness in response to an accumulating succession of said measures.

35. In imaging apparatus according to claim 34, the further improvement in which said measure-providing means includes means for providing a unity ratio in response to radiance values within a selected measure of one another.

36. In imaging apparatus according to claim 34 the further improvement in which said measure-providing means includes means for producing said measure, for each second area which is paired with a first area located beyond the image field, in response exclusively to the measure already assigned to that second area.

37. In imaging apparatus according to claim 30, the further improvement in which A. each segmental area corresponds to a coordinate-identified location of said image field, and B. said means for pairing provides a pairing of each of plural second areas with a different first area removed therefrom by the same coordinate direction and spacing.

38. In imaging apparatus according to claim 37, the further improvement A. in which said means for pairing segmental areas includes means for providing said number of pairs sequentially, B. comprising means for assigning each segmental area an initializing prior measure and for updating each prior measure in response to the measure provided for each sequential pairing, and C. comprising means for providing said measure, for each second area paired with a first area located beyond said image field in said coordinate direction, in response only to the prior measure assigned to that second area.

39. Image processing apparatus comprising A. first and second signal adder means, B. first and second delay means, each arranged to apply signals output therefrom to an input of the same-numbered adder means, C. first and second signal transformation means, each arranged to apply signals output therefrom to a further input of the same-numbered adder means, said first transformation means having a polarity inversion function and said second transformation means having a reset function and being arranged to receive signals output from said first adder means, D. first memory means arranged to apply signals read therefrom to said first transformation means and to said first delay means, E. second memory means arranged to apply signals read therefrom to a further input of said first adder means and to said second delay means, F. further signal transformation means having a compress function and arranged to receive signals from said second adder means and to apply signals to an input of said second memory means, and G. control means for controlling said adder means, delay means, transformation means and memory means for applying signals to said first adder means from said first transformation means and from said first delay means and from said second memory means with selected relative timing, and to apply signals to said second adder means from said second transformation means and from said second delay means with selected timing relative to one another and relative to said application of signals to said first adder means.

40. Lightness-imaging apparatus having means for providing information identifying optical radiance associated with arrayed sections of a selected image field, said apparatus further comprising A. means for pairing identically configured and sized segmental areas of said viewing field differently a number of times and for providing a multiple of sets of said different pairings, each said set involving areas of a size different from other sets, B. means for providing, for each pairing of segmental areas, a comparative measure of said radiance information at the paired areas, C. means for resetting each said measure with reference to a selected limit, and D. means for determining image lightness for each arrayed section of the image field in response to a plurality of said reset measures.

41. Imaging apparatus according to claim 40 A. further comprising means for assigning each segmental area an initializing value of said measure, B. in which said measure-providing means includes (1) means for providing each said measure in response to the product of a ratio function of the radiance information associated with the two paired areas and the measure already assigned to a first of said paired areas and (2) means for replacing the measure assigned to each second area of a pair in response to said product, and C. in which said means for determining includes (1) means for providing said replaced measures for each set of pairings sequentially for different sets thereof, and (2) means for producing an initializing value of said measure for all but the sequentially first set of pairings in response to the replaced measure produced with the last pairing of the sequentially preceding set thereof.

42. Imaging apparatus according to claim 40 A. in which said measure-providing means includes means for providing multiple measures of image field lightness for said pairings in each set thereof and B. in which said means for determining includes means for arithmetically combining said measures from each set of pairings.

43. Lightness-imaging apparatus having means for providing information identifying optical radiance associated with each arrayed section of a selected image field said apparatus further comprising A. means for selectively pairing segmental areas of said image field a selected number of times, each said pairing being of segmental areas of identical configuration and size, B. means for providing, for each pairing of segmental areas, at least one measure of transition in said radiance information between the paired areas, said measure conforming to the equation:

log ip(xy) = log op(o,o) + log r(x,y) - log r(o,o) where log ip(x,y) is the log of the measure for a first segmental area at location (x,y) in the image field relative to a reference location for a second area paired therewith, log op(o,o) is the log of the measure previously assigned to or determined for said second segmental area at said reference location (o,o) in the image field and paired with said first area, and log r(x,y) and log r(o,o) are the logs of the radiance information for said first and second paired areas, respectively, C. means for resetting each said measure with reference to a selected limit, and D. means for determining image lightness for each arrayed section of the image field in response to an arithmetic averaging of a plurality of said reset measures, at least some of which are provided for pairings which differ from one another in at least one spatial parameter.

44. Lightness-imaging apparatus having means for providing information identifying optical radiance associated with each arrayed section of a selected image field, said apparatus further comprising A. means for selectively pairing segmental areas of said image field a selected number of times, each said pairing being of segmental areas of identical configuration and size, B. means for providing, for each pairing of segmental areas, at least one measure of transition in said radiance information between the paired areas, said measure conforming to the equation:

log np(x,y) =

1/2(log op(x,y)] + [log op (o,o)+log r(x,y)-log r(o,o)]) where log np(x,y) is the log of the measure for a first segmental area at location (x,y) in the image field relative to a reference location for a second area paired therewith, log op(x,y) is the log of the measure previously assigned to or determined for said first area, log op(o,o) is the log of the measure previously assigned to or determined for said second segmental area at said reference location (o,o) in the image field and paired with said first area, and log r(x,y) and log r(o,o) are the logs of the radiance information for said first and second paired areas, respectively, and where each said term [log op (o,o) + log r(x,y)-log r(o,o)] is reset with reference to a selected limit.

45. Imaging apparatus according to claim 44 further comprising means for determining image lightness for each arrayed section of the image field in response to an arithmetic averaging of a plurality of said reset measures, at least some of which are pro-vided for pairings which differ from one another in at least one spatial parameter.

46. A method for producing an image of a subject comprising the steps of A. detecting radiance ratios between different areas of said subject and producing a first lightness-determining quan-tity in response to each such ratio, B. effecting said ratio detection for each area of said subject a multiple number of times with other areas of said sub-ject which are at different locations on said subject relative to that area, C. combining each first lightness-determining quantity with a second lightness-determining quantity associated with one area in that ratio and replacing the second lightness-determining quantity associated with another area in that ratio in response thereto, and D. producing an image of the subject in which the light-ness of the respective image areas is determined by the last re-placement values of said second lightness-determining quantities.

47. An image-processing method for determining a field of accumulating measures of image lightness in response to information identifying optical radiance associated with an image field, said method having the improvement comprising the steps of A. sequentially determining a comparative measure of the radiance information for each segmental area of said image field relative to said information for each of plural other seg-mental areas, said method-determining step including (i) providing a new intermediate value of each such measure in response to the product of a ratio function of the radiance information associated with each first-named segmental area and with each second-named segmental area and of a like mea-sure previously determined for the second-named segmental area, (ii) determining a sequentially new value of each said measure in response to a selectively weighted averaging of said new intermediate value and a like measure previously deter-mined for said first-named segmental area, and B. updating the previously determined measure for each first-named segmental area in response to said sequentially-deter-mined new value, thereby to determine each measure in the field in response to an accumulating succession of said measures.

48. An image processing method according to claim 47 further characterized in that said measure-determining step includes determining different ones of each said plurality of measures for first-named and second-named areas that correspond to spatially different sections of said image field, where said spatial dif-ference includes at least a difference in size of or in separation between first-named and second-named areas for which that measure is being determined.

49. An image processing method for determining a field of accumulating measures of image lightness in response to informa-tion identifying optical radiance associated with an image field, said method having the improvement comprising the steps of A. sequentially determining a comparative measure of the radiance information for each segmental area of said image field relative to said information for each of plural other seg-mental areas of that field, said measure-determination including (i) providing a new intermediate value of each such comparative measure in response to the product of a ratio function of the radiance information associated with each first-named seg-mental area and with each second-named segmental area and of a like measure previously determined for the second-named segmental area, and further providing each such intermediate value with reference to a selected condition of said product, and (ii) determining a sequentially new value of each said comparative measure in response to a selectively weighted averaging of said new intermediate value and a like measure pre-viously determined for said first-named segmental area, and B. replacing the previously-determined measure for each first-named segmental area in response to said sequentially new value, thereby to determine each measure in the field thereof in response to an accumulating succession of said comparative measures.

50. An image processing method according to claim 49 further characterized in that said measure-determining step includes de-termining said measure for each first-named area relative to each of a set of second-named other areas, where said set includes areas at selected different image-field locations relative to said first-named area.

51. An image processing method according to claim 49 further characterized in that said measure determining step includes effec-ting a determination of said measure for each of a selected plur-ality of areas prior to effecting a further such determination for an area of that plurality.

52. An image processing method for determining information corresponding to image lightness in response to radiance-identify-ing information for a selected image field, said method having the improvement comprising the steps of A. representing said radiance-identifying information for each of selected segmental areas of the viewing field, B. determining a selected comparison measure between said identifying information for each segmental area and said information for another segmental area, and determining therefrom and from a previously-determined lightness-identifying quantity for each latter segmental area a newly-determined lightness-identifying quantity for each former segmental area, C. effecting a selected multiple of said computations sequentially between segmental areas that correspond to differently-spaced locations in said field of view, and D. producing said lightness-information for said image field in response to said multiple determinations.

53. An image processing method according to claim 52 having the further improvement in which said step of effecting said determinations sequentially includes applying each newly-determined quantity of one determination as a previously-determined quantity in a subsequent determination.

54. An image processing method according to claim 52 having the further improvement in which said step of effecting said determinations with differently-spaced locations includes ordering said determinations according to the magnitude of the spacing between locations.

55. An image processing method according to claim 52 having the further improvement in which said step of determining includes operating on a set of said identifying information that is responsive to said selected image field to effect one said determination for each segmental area of the image field prior to effecting a further determination for any such segmental area.

56. An image processing method according to claim 52 having the further improvement in which said step of effecting multiple determinations is further adapted for producing for each segmental area a lightness-identifying quantity that is responsive to rad-iance-identifying information for substantially every other seg-mental area of the image field.

57. An image processing method for determining information corresponding to image lightness in response to radiance-identify-ing information for a selected image field, said method having the improvement comprising the steps of A. representing said radiance-identifying information for each of selected segmental areas of the viewing field, B. determining a selected comparison measure between said identifying information for each segmental area and said information for another segmental area, and determining therefrom and from a previously-determined lightness-identifying quantity for each former segmental area a newly-determined lightness-iden-tifying quantity for each latter segmental area, said measure-determination including means for effecting a determination of said measure for each of a selected plurality of areas prior to effecting a further such determination for an area of that plur-ality, C. effecting a selected multiple of said de-terminations for said plurality of areas sequentially, different ones of said sequential determinations being between segmental areas having a spatial parameter different from that of the areas of other such determinations, and applying each newly-determined quantity of one determination as a previously-determined quantity in a subsequent determination, and further including ordering said determinations according to the magnitude of said spatial parameter, and D. producing said lightness-information for said image field in response to said multiple determinations.

58. A lightness-imaging method in which information is provided identifying optical radiance associated with arrayed sections of a selected image field, said method further comprising the steps of A. selectively grouping segmental areas of said image field a selected number of times, different ones of at least some of said groupings involving areas having at least one spatial parameter different from other groupings of areas, B. providing, for each grouping of segmental areas, at least one measure of visually significant transition in said radiance information between areas of that grouping, said measures being with reference to a selected lightness condition, and C. determining image lightness for each arrayed section or the image field in response to a plurality of said measures, at least some of which are provided for groupings which differ from one another in at least one spatial parameter selected from the parameters of distance, direction and size.

59. An imaging method according to claim 58 in which said grouping step provides at least a first of said groupings with areas of uniform size and provides different ones of such groupings among areas having different separations in terms of at least one spatial parameter selected from the parameters of distance and of direction.

60. An imaging method according to claim 58 A. further comprising the step of assigning each segmental area an initializing value of said measure, B. in which said measure-providing step includes provid-ing each said measure in response to the product of a ratio func-tion of the radiance information associated with grouped areas and the measure already assigned to a first of said grouped areas, and C. further comprising the step of assigning a replacement value to the measure assigned to a second of said grouped areas in response to said product-responsive measure.

61. An imaging method according to claim 60 further including the step of resetting said product-responsive measure with refer-ence to said selected lightness condition.

62. An imaging method according to claim 60 in which said measure-providing step includes imposing a threshold on said ratio function.

63. An imaging method according to claim 60 further comprising the step of retaining the measure assigned to said second of said grouped areas, in lieu of said assignment of a replacement value for each said second grouped area which is paired with an area located beyond said image field.

64. An imaging method according to claim 58 in which said determining step includes determining image lightness in response to an arithmetic averaging function of plural ones of said measures provided for groupings which differ from one another in at least one spatial parameter.

65. An imaging method according to claim 58 in which said measure-providing step includes providing at least some said measures in sequence with one another and for providing sequentially-successive measures in response, at least partially, to a preceding measure for an area of a grouping, thereby to determine said image lightness in response to an accumulating succession of said measures.

66. An imaging method according to claim 65 A. in which said grouping step provides groupings which differ by a magnitude parameter, and B. in which said measure-providing step includes providing said sequential measures ordered between groupings of largest spatial parameter and groupings of smallest parameter.

67. In a lightness-imaging method in which information is provided identifying optical radiance associated with each arrayed section of a selected image field, and including (i) selectively pairing segmental areas of said image field a selected number of times, each said pairing being of segmental areas of identical configuration and size, (ii) providing, for each pairing of segmental areas, at least one comparative measure of said radiance information at the paired areas, and (iii) resetting each said measure with reference to a selected limit condition, the improvement comprising the further step of determining image lightness for each arrayed section of the image field in response to a plurality of said reset measures, at least some of which are provided for pairings which differ from one another in at least one spatial parameter.

68. In an imaging method according to claim 67, the further improvement A. comprising the step of assigning each segmental area an initializing value of said measure, B. in which said measure-providing step includes providing each said measure in response to the product of a ratio function of the radiance information associated with paired areas and the measure already assigned to a first of said paired areas, and C. comprising the step of assigning a replacement value to the measure assigned to a second of said paired areas in res-ponse to said product-responsive measure.

69. In an imaging method according to claim 67, the further improvement in which said step of pairing segmental areas includes providing different ones of said pairs of said areas with differ-ent spacing parameters ordered in said sequence from pairings of the largest parameter to pairings of the smallest parameter.

70. In an imaging method according to claim 67, the further improvement A. in which said step of pairing segmental areas includes providing said number of pairs sequentially, B. comprising the step of assigning each segmental area an initializing value of said measure, and C. in which said measure-providing step includes (1) providing an intermediate value of each said measure in response to the product of a ratio function of the radiance information associated with paired areas and the measure already assigned to a first area of each pair, (2) providing each said measure in response to a selectively weighted averaging of the last previously provided intermediate value for said first area of each pair and the last previously provided prior measure for a second area of each pair, and (3) providing a replacement value for the prior measure for each second area in response to said averaging, there-by to determine image lightness in response to an accumulating succession of said measures.

71. In an imaging method according to claim 70, the further improvement in which said measure-providing step includes providing a unity ratio in response to radiance values within a selected measure of one another.

72. In an imaging method according to claim 70, the further improvement in which said measure-providing step includes producing said measure, for each second area which is paired with a first area located beyond the image field, in response exclusively to the measure already assigned to that second area.

73. In an imaging method according to claim 67 in which each segmental area corresponds to a coordinate-identified location of said image field, the further improvement A. in which said pairing step provides a pairing of each of plural second areas with a different first area removed therefrom by the same coordinate direction and spacing, B. in which said step of pairing segmental areas includes providing said number of pairs sequentially, and C. comprising the steps of (1) assigning each segmental area an initializing prior measure, (2) updating each prior measure in response to the measure provided for each sequential pairing, and (3) providing said measure, for each second area paired with a first area located beyond said image field in said coordinate direction, in response only to the prior measure assigned to that second area.

74. A lightness-imaging method in which information is provided identifying optical radiance associated with arrayed sections of a selected image field, said method comprising the step of A. pairing identically configured and sized segmental areas of said viewing field differently a number of times and for providing a multiple of sets of said different pairings, each said set involving areas of a size different from other sets, R. providing, for each pairing of segmental areas, a comparative measure of said radiance information at the paired areas, C. resetting each said measure with reference to a selected limit condition, and D. determining image lightness for each arrayed section of the image field in response to a plurality of said reset measures.

75. An imaging method according to claim 74 A. further comprising the step of assigning each segmental area an initializing value of said measure, B. in which said measure-providing step includes (1) providing each said measure in response to the product of a ratio function of the radiance information associated with the two paired areas and the measure already assigned to a first of said paired areas, and (2) replacing the measure assigned to each second area of a pair in response to said product, and C. in which said determining step includes (1) providing said replaced measures for each set of pairings sequentially for different sets thereof, and (2) producing an initializing value of said measure for all but the sequentially first set of pairings in response to the replaced measure produced with the last pairing of the sequentially preceding set thereof.

76. A liahtness-imaging method in which information is provided identifying optical radiance associated with each arrayed section of a selected image field, said method comprising the steps of A. selectively pairing segmental areas of said image field a selected number of times, each said pairing being of segmental areas of identical configuration and size, B. providing, for each pairing of segmental areas, at least one measure of transition in said radiance information between the paired areas, said measure conforming to the equation:

log ip(xy) = log op(o,o) + log r(x,y) - log r(o,o) where log ip(x,y) is the log of the measure for a first segmental area at location (x,y) in the image field relative to a reference location for a second area paired therewith, log op(o,o) is the log of the measure previously assigned to or determined for said second segmental area at said reference location (o,o) in the image field and paired with said first area, and log r(x,y) and log r(o,o) are the logs of the radiance information for said first and second paired areas, respectively, C. resetting each said measure with reference to a selected limit, and D. determining image lightness for each arrayed section of the image field in response to an arithmetic averaging of a plurality of said reset measures, at least some of which are provided for pairings which differ from one another in at least one spatial parameter.

77. A lightness-imaging method in which information is provided identifying optical radiance associated with each arrayed section of a selected image field, said method comprising the steps of A. selectively pairing segmental areas of said image field a selected number of times, each said pairing being of segmental areas of identical configuration and size, B. providing, for each pairing of segmental areas, at least one measure of transition in said radiance information between the paired areas, said measure conforming to the equation:

log np(x,y) =

1/2{[log op(x,y)] + [log op (o,o)+log r(x,y)-log r(o,o)]}

where log np(x,y) is the log of the measure for a first segmental area at location (x,y) in the image field relative to a reference location for a second area paired therewith, log op(x,y) is the log of the measure previously assigned to or determined for said first area, log op(o,o) is the log of the measure previously assigned to or determined for said second segmental area at said reference location (o,o) in the image field and paired with said first area, and log r(x,y) and log r(o,o) are the logs of the radiance information for said first and second paired areas, respectively, and where each said term [log op (o,o) +
log r(x,y) log r(o,o)] is reset with reference to a selected limit.

78. An imaging method according to claim 77 further comprising the step of determining image lightness for each arrayed section of the image field in response to an arithmetic averaging of a plurality of said reset measures, at least some of which are provided for pairings which differ from one another in at least one spatial parameter.

79. Image processing apparatus comprising A. means for receiving information responsive to the radiance values defining an image field, and B. means for deriving from said information a lightness field containing final lightness values for predetermined segmental areas of said image field, said final lightness value deriving means establishing initial lightness values for all areas of said image field and sequentially performing a selected number of pro-cess steps for said image field, in each step of which process selected areas of said image field are selectively paired with different areas of said image field and in successive steps of which process such pairings of areas differ from other pairings in at least one spatial parameter according to a predetermined sequence,and in each of which steps such paired areas are compared to establish a new lightness value for each said selected area as a function of the ratio of its radiance value to that of the different area with which it is paired and as a function of light-ness values established for such paired areas in a preceding pro-cess step, and wherein said final lightness value for each said segmental area comprises an effective comparison of information responsive to its radiance value to information responsive to the radiance value from substantially all other areas of said image field without a direct comparison to each of said other segmental areas.

80. Image processing apparatus according to claim 79 in which said lightness-value deriving means includes means for establishing each new lightness value with reference to at least one selected lightness condition.

81. Image processing apparatus according to claim 79 in which said lightness value deriving means includes means for pairing areas of uniform like size throughout at least a selected portion of said process steps and for selecting at least one spatial para-meter for pairing, in successive steps, areas spaced apart by a distance that decreases progressively in the course of at least said portion of said process steps.

82. Image processing apparatus according to claim 79 in which said lightness value deriving means includes means for pairing areas of like size in each said process step, and for selecting at least one spatial parameter to decrease the sizes, in said image field, of said paired areas in at least selected different steps progressively in the course of said process steps.

83. Image processing apparatus according to claim 79 in which said lightness value deriving means includes means for selecting said one spatial parameter to decrease in magnitude, at at least selected successive steps, in the course of said process steps.

84. Image processing apparatus comprising A. means for receiving information responsive to the radiance values defining an image field, and B. means for deriving from said information a lightness field containing final lightness values for predetermined segmental areas of said image field, said final lightness value deriving means establishing initial lightness values for all areas of said image field and performing a selected number of process steps for said image field, in each step of which process selected areas of said image field are selectively paired with different areas of said image field and in different steps of which process such pairings of areas differ selectively from other pairings in at least one spatial parameter, and in each of which steps such paired areas are compared to establish a new lightness value for each said selected area as a function of the ratio of its radiance value to that of the different area with which it is paired and as a function of lightness values established for such paired areas in a different process step, and in which said one spatial parameter is selected, for at least selected different ones of such pairings, to establish new lightness values for areas that are at least relatively small or relatively closely spaced apart using lightness values established for areas that are compara-tively larger or comparatively further spaced apart, and wherein said final lightness value for each said segmental area comprises an effective comparison of information responsive to its radiance value to information responsive to the radiance value from sub-stantially all other areas of said image field without a direct comparison to each of said other segmental areas.

85. An image processing method comprising the steps of A. receiving information responsive to the radiance values defining an image field, and B. deriving from said information a lightness field con-taining final lightness values for predetermined segmental areas of said image field, said final lightness value deriving step establishing initial lightness values for all areas of said image field and sequentially performing a selected number of process steps for said image field, in each step of which process selected areas of said image field are selectively paired with different areas of said image field and in successive steps of which process such pairings of areas differ from other pairings in at least one spatial parameter according to a predetermined sequence, and in each of which steps such paired areas are compared to establish a new lightness value for each said selected area as a function of the ratio of its radiance value to that of the different area with which it is paired and as a function of lightness values established for such paired areas in a preceding process step, and wherein said final lightness value for each said segmental area comprises an effective comparison of information responsive to its radiance value to information responsive to the radiance value from substantially all other areas of said image field with-out a direct comparison to each of said other segmental areas.

86. An image processing method comprising the steps of A. receiving information responsive to the radiance values defining an image field, and B. deriving from said information a lightness field con-taining final lightness values for predetermined segmental areas of said image field, said final lightness value deriving step establishing initial lightness values for all areas of said image field and performing a selected number of process steps for said image field, in each step of which process selected areas of said image field are selectively paired with different areas of said image field and in different steps of which process such pairings of areas differ selectively from other pairings in at least one spatial parameter, and in each of which steps such paired areas are compared to establish a new lightness value for each said selected area as a function of the ratio of its radiance value to that of the different area with which it is paired and as a function of lightness values established for such paired areas in a different process step, and in which said one spatial para-meter is selected, for at least selected different ones of such pairings, to establish new lightness values for areas that are at least relatively small or relatively closely spaced apart using lightness values established for areas that are comparatively larger or comparatively further spaced apart, and wherein said final lightness value for each said segmental area comprises an effective comparison of information responsive to its radiance value to information responsive to the radiance value from substan-tially all other areas of said image field without a direct com-parison to each of said other segmental areas.